I can't get docker to work (on Ubuntu 20.04). The curl install does not work. But the "sudo apt install..." did work. I had to add "sudo" to the "./start.sh" command. It down loaded the docker container. I can't run the "roslaunch ... ...". It complains about networking. Someone had asked that on ROS Agriculture Slack, but that has all been deleted. Even if that worked, I see there are apparently two different ways to start a docker package...

So I decided to go back to what I know. I loaded VirtualBox on Ubuntu 20.04. I created a Ubuntu 16.04 virtual machine. I am loading ROS kinetic on that. Then I will clone which ever respository seems appropriate. Maybe I will clone the Ubuntu/Kinetic machine and have both simulators loaded that I can choose from.

*Thread Reply:* Q: Re: networking error,

$ docker run -p 6080:80 --network=”host” --device=/dev/input/js0 rosagriculture/lawn_tractor

I had trouble with the above statement and removed the '--network=”host”' portion

Q: Are you following someone's instructions? It would be interesting to see how they have documented their steps.

*Thread Reply:* hi @JeffS

If you would like to use Docker as a non-root user, you should now consider adding your user to the “docker” group with something like: sudo usermod -aG docker <your-user> Remember to log out and back in for this to take effect!

look the sreencap or the link..

https://docs.docker.com/engine/install/ubuntu/

I got VirtualBox loaded on 20.04. I created an image with Ubuntu 16.04 and ROS Kinetic. I will follow the Docker instrucions again. Here: https://github.com/ros-agriculture/lawn_tractor/blob/master/simulator.md I will see if these instructions work under 16.04. (Mainly as an experiment.)

But, I don't plan to put much effort into Docker.

I ordered a new hard drive yesterday. I plan to swap the 20.04/Noetic/Foxy hard drive in my robot laptop with the new one and build a fresh install of 16.04/Kinetic. Then clone the actual repositories and run it without virtual machines or Docker. I'll let you know when I have that loaded.

GitHub

ros-agriculture/lawn_tractor

Software for self driving lawn tractor. Contribute to ros-agriculture/lawn_tractor development by creating an account on GitHub.

Original URL: https://github.com/ros-agriculture/lawn_tractor/blob/master/simulator.md

I can run a waypoint mission around the green box in the simulator. The waypoints are referenced at the bottom of this file in the main Google folder.

I loaded Docker from this page: https://github.com/ros-agriculture/lawn_tractor/blob/master/simulator.md on 16.04, under VirtualBox, under 20.04.

The curl command works under 16.04, but did not work under 20.04. If you load it with curl it gives you the message about: sudo usermod -aG docker <your-user> I didn't see that when I loaded using the "sudo app install" method. I may have just missed it.

I still had to add sudo to the ./start.sh command. Looks like somebody did something cute inside the start.sh script.

When I did the roslaunch... it took about 10 minutes before ROS time started counting and the tractor turned red. Then I could select a destination and it would drive. But all of the other windows were running real slow and the whole thing was un-usable.

So I loaded the "Local" version from the install page (with 16.04 under VirtualBox under 20.04). That seemed to work and behave much better.

So tomorrow when I get my new drive I will load that and hopefully have a native install that works well enough to do development.

GitHub

ros-agriculture/lawn_tractor

Software for self driving lawn tractor. Contribute to ros-agriculture/lawn_tractor development by creating an account on GitHub.

Original URL: https://github.com/ros-agriculture/lawn_tractor/blob/master/simulator.md

I just started it again. It still takes 3-1/2 minutes to start up. (the local install under the virtual machine) So hopefully a fresh install with native Ubuntu 16.04 (on a SSD) will improve it.

Well at least we are all near the same place. I'm working in the ros in spanish community.. we are just working in the translation of the different type of install scripts. There are a lot of options but One thing that makes all this very confusing to understand is which problems are yours alone and which are common to everybody. The situation that you suffer with the installation is na example of that. I'm working in documenting all types of installations..

My new (cheap, 250GB SSD) showed up. In two hours I had it installed, loaded Ubuntu 16.04, ROS Kinetic, lawn_tractor simulator (local version).

It works. It boots from power-off in 50 seconds (due to SSD). When I run roslaunch lawn_tractor_sim lawn_tractor_sim.launch it takes about 15 seconds until it is driveable.

I'm pretty happy with this setup. My 20.04 setup is slow. Everything about it is slow. I may have to down grade to something else if I can't figure out what is wrong with it...

*Thread Reply:* I have a

*Thread Reply:* I bought it used off of ebay

*Thread Reply:* Juan said: "So you are running ROS kinetic with ubuntu 16.04.. That will be your final setup..? Because if that's right we can began loading it to the spreadsheet.." (I don't know hot to do a quote...

That is my final setup for today. I'm not really interested in updating the spread or even being in the spread sheet.

*Thread Reply:* I am running 16.04/Kinetic. I don't know about "final". At the moment I don't have a need to change so I guess I will stay where I'm at until there is a reason to switch. I'll use whatever works reliably and has the most support.

[Does anybody know how to prevent Slack from appending messages togher? Or do I just have to wait 10 minutes before posting another topic?]

I have a thought on joysticks. Both of my computers install the joystick as js1 and not js0. So I always have to hack the config files.

What if a guy were to create an alias for a particular computer? Maybe alias js0 (or js1) to a name, say, JOYSTICKALIAS. That could be done in the .bashrc file. Then edit all of the config files to use JOYSTICKALIAS instead of js0 (or js1).

*Thread Reply:* Naaa, I hate slack, it makes me fell old.. but seems to be the only tool posible.. 🤷‍♂️

*Thread Reply:* I don't know about appending messages together. In a channel I think you can only reply to a thread or at the bottom start a new thread.

*Thread Reply:* No, this is what I'm talking about. Final thought...

*Thread Reply:* Oh, I have a new thought. But look... it appends it to the previous message...

*Thread Reply:* Oh course, now it didn't. Maybe it is context sensitive. This was in a thread. Let me go try it in the main channel.

*Thread Reply:* for me it is incomprehensible and the only way to see what happens is to see the threads of the conversations; but like that, several times I lost the context and order of the conversations

Oh, the alias thing doesn't fix my problems. Alias seems to work for commands but not for variables...

A bigger problem is that my joystick shows up as js0 or js1, depending if the joystick is plugged in when I turn the computer on.

I should look at udev rules to identify my Logitech gamepad. Then create a device name like "JOYSTICK". Then use that as /dev/input/JOYSTICK instead of /dev/input/js0.

(example) This is what I'm talking about. Final thought...

*Thread Reply:* You have started a new message. I am replying to the thread.

This is a totally new thought... Does it get appended or does it start a new message?

*Thread Reply:* It is a new thought/subject.

*Thread Reply:* I would avoid checking the box at the bottom, regarding "Also send to". That seems to be duplicating messages.

*Thread Reply:* Well that is more Slack non-sense. If you don't check the box it is only appened to the thread. A user will be notified if "they are following the thread" (whatever that means). But if you check the box then it gets added (as a duplicate to the main channel) and everybody gets notified. I personally would not use threads since it just adds to the confusion, but that is the way the Slack is written. But if someones starts a thread then I will reply to the thread if I want to answer someone.

*Thread Reply:* The bottom line is... Slack is free and it mostly works. So I am willing to put up with it.

Now if I wait 10 minutes (or maybe 3 or 5 minutes it will start a new message. Or if somebody else posts before I do.

So the moral of the story is... You just have to guess at what the results will be. I'm assuming the people that write the Slack software don't actually use it...

Just a suggestion:

I'm looking at this file: https://docs.google.com/document/d/1uXMrZL0K-PJ7fLSx8w51RZotoYZD70tybRhmNpjhyyU/edit Named: Using Matt's waypoint publisher

It would be real handy to have a synopsis at the top. Maybe something like this:

Waypoint Publisher This will read waypoints from a file and send them to ... (movebase_flex?) to drive a path. This consists of two python scripts and a text file containing a list of waypoints.

I extracted this evaluation from digging through your steps.

I assume if I build the python scripts and create a waypoint file that I should be able to follow a path without using Docker or VS-Code or Terminator. But I will have to try it to see...

*Thread Reply:* I can add it or you can add it; You should have edit rights. Re: "without using Docker...", I assume so. It is simply meant to help people go fast; Not a hard requirement.

Why does Slackbot keep suugesting that I load github app and Google Drive Link app? Do I need those? What would they do to help me?

*Thread Reply:* No idea.

fyi, looks interesting to me - I've spoken to him before about RTK....https://deepsouthrobotics.com/2019/08/23/anatomy-of-a-huge-self-driving-mower/

I got my tractor to follow the path as Al described in his instructions. I read through his instructions and pulled out the pieces I thought I needed. I created a package named "jeff" to put my new files in. So I put the files here: ~/catkin_ws/src/jeff/python/runmission$ The 3 files are: generated_points.txt path_follower_node.py pathpublisher.py I noticed several things. I have to execute the ~/catkin_ws/src/jeff/python/runmission$ rosrun jeff pathpublisher.py from the runmission directory. Otherwise the pathpublisher does not find the generated_points.txt file. I will have to figure that out. It does not start moving on the first path generated. It starts moving when the second path is published. It generates more paths while it is moving but they are ignored. When it finishes and I get "GOAL Reached!" and it stops. It will not start again until I shut down the simulator launch file and restart it. So I will have to figure that out. It doesn't drive very smoothly and puts out warnings. But it goes where it is supposed to. That is probably tuning... I also notice the packages I downloaded using git clone don't show up as packages. (when I do rospack find ...) So I should look at that. But since I got it move I now have things to look at...

*Thread Reply:* "...It will not start again until I shut down the simulator launch file and restart it...., I have the same experience. I'm guessing there are some path cancel commands needed.; "...It doesn't drive very smoothly...", I have some notes in this file. Basically I think I learned the system wants a point at least every meter.

I got Gazebo running and I got a random Ackermann vehicle to run.

*Thread Reply:* Just posting this here for future reference https://github.com/carla-simulator/ros-bridge/blob/master/carla_ackermann_control/README.md

GitHub

carla-simulator/ros-bridge

ROS bridge for CARLA Simulator. Contribute to carla-simulator/ros-bridge development by creating an account on GitHub.

Original URL: https://github.com/carla-simulator/ros-bridge/blob/master/carla_ackermann_control/README.md

*Thread Reply:* This is newer than what I found. And it is not an end-less derivation of this original I found: https://github.com/jbpassot/ackermann_vehicle This is the newest version of that I found: https://gitlab.yonohub.com/mohamedahmed/picar/tree/master/src Which is the one with red wheels.

FYI: I am not suggesting that anyone else go down the Gazebo path. Maybe if I get something usable...

GitHub

jbpassot/ackermann_vehicle

ROS packages for simulating a vehicle with Ackermann steering - jbpassot/ackermann_vehicle

Original URL: https://github.com/jbpassot/ackermann_vehicle

GitLab

src · master · Mohamed Hassan / PICAR

GitLab Community Edition

Original URL: https://gitlab.yonohub.com/mohamedahmed/picar/tree/master/src

Nice

Throu cmdvel..?

That is not using cmdvel. It is using ackermanncmd. (as shown in the image.)

BTW, if you try to run roslawntractor/lawntractornavigation/launch/sendgoals.launch you will find that the file "followgps_waypoints" is missing. I searched the original Slack channel and did not find it.

*Thread Reply:* I thought maybe the docker image did not match the main repository, but they do match and it is missing in both places.

I did not try to upgrade the docker. I see there is a page that suggests that may be possible: https://hub.docker.com/u/rosagriculture

I don't know if it is newer or not.

*Thread Reply:* https://drive.google.com/drive/u/0/folders/1KBipHdNtvZFWpMKnAJSvnViyiNH2jg1k

*Thread Reply:* I'm not sure it's going to work but I'm uploading everything I have so you can see it.

*Thread Reply:* Maybe I found the solution. This video says to gpsgoalserver: https://www.youtube.com/watch?v=lTqUMAUM1cc https://github.com/ros-agriculture/gps_goal_server I'll check it out...

YouTube

} ROS Agriculture (https://www.youtube.com/channel/UCnr6AObsHnHSGPge0HVlBpA)

Lawn tractor meeting 12052020

Original URL: https://www.youtube.com/watch?v=lTqUMAUM1cc

GitHub

ros-agriculture/gps_goal_server

Set a ROS navigation goal using latitude and longitude. - ros-agriculture/gps_goal_server

Original URL: https://github.com/ros-agriculture/gps_goal_server

*Thread Reply:* Hey, this a perfect time to use my video index I am creating: 12052020 Lawn tractor meeting https://www.youtube.com/watch?v=lTqUMAUM1cc 00:00 Juan discusses documentation 12:30 Al talking about adding satellite image to RVIZ. 15:20 Johanthan announces a video presentation 20:20 Doug has a new Oculus Quest 2 headset. Mentions microcontrollers. 23:55 Matt asks if people are having problems with Docker networking. Lots of discussion about Docker and the simulator. 41:10 Al requests a model for an Ackermann vehicle. 45:50 Jeff asks about Matt's drone.

*Thread Reply:* I didn't hear a mention of gpsgoalserver in the video. But there is a link in the comments below the video.

Might be some interesting code here: https://hackaday.io/project/25406-wild-thumper-based-ros-robot I cannot open the git repository referenced from work.

hackaday.io

Wild Thumper based ROS robot

I've build a few robots before, but never was able to do real navigation with them, e.g. drive successfully from living room to kitchen. Reason was that a few infrared and sonar sensors are not enough to do 'real' navigation. You can avoid obstacles with them, but not much more. My Software was not much better suited for this until I discovered ROS (Robot Operating System) with the rviz GUI a few years ago. Since ready available robots with ROS are a little bit expensive I decided to build a new robot in spring 2015 around the wild thumper 4wd chassis. The hardware and electronics were almost completed in a few weeks. Since then I've spent some time with tuning.. The goal is to be able to navigate indoor doing SLAM (simultaneous localization and mapping) and outdoor with GPS. The free space on the back is supposed to host a robot arm somewhere in the future.

Original URL: https://hackaday.io/project/25406-wild-thumper-based-ros-robot

*Thread Reply:* This does look interesting. Worth a look.

*Thread Reply:* He does have a gpsfollowwaypoints.py script. Looks like it reads waypoints from a file, sends them to move_base and waits for each one to complete.

*Thread Reply:* Looks like he has a movebase (not movebase_flex) version of a gps waypoint module.

*Thread Reply:* $ git clone <https://defiant.homedns.org/git/ros_wild_thumper.git>

Al, are any of these towns close to you in Texas?

jeff@jeff-hp-pavilion:~$ grep -i tx rtk2gosourcetable.txt STR;AUSDUVAL;Austin, TX;RTCM 3.2;1005(1),1074(1),1084(1),1124(1),1230(1);0;GPS+GLO+BDS;SNIP;USA;30.30;-97.73;1;0;sNTRIP;none;N;N;0;none; STR;AUSLOFTTX;Austin, Tx;RTCM 3.2;1005(1),1074(1),1077(1),1084(1),1087(1),1094(1),1097(1),1124(1),1127(1),1230(1),4072(1);;GPS+GLO+GAL+BDS;SNIP;USA;30.28;-97.69;1;0;sNTRIP;none;N;N;10860;none; STR;EPCWID-Fabens;Fabens, TX;RTCM 3.1;1004(1),1006(10),1008(10),1012(1),1033(10),4091(10);2;GPS+GLO;SNIP;USA;31.50;-106.16;1;0;sNTRIP;none;N;N;1940;none; STR;PROVEN;Wichita Falls, Tx;RTCM 3.1;1004(1),1006(10),1008(10),1012(1),1033(10),4091(10);2;GPS+GLO;SNIP;USA;33.89;-98.53;1;0;sNTRIP;none;N;N;2540;; STR;VN1-legacy;Dallas, Tx;RTCM 3.2;1004(1),1005(1),1012(1),1033(1),1230(1);2;GPS+GLO;SNIP;USA;32.89;-96.70;1;0;sNTRIP;none;N;N;3140;none; STR;VN1;Dallas, Tx;RTCM 3.2;1005(1),1077(1),1087(1),1097(1),1127(1),1230(1);;GPS+GLO+GAL+BDS;SNIP;USA;32.89;-96.70;1;0;sNTRIP;none;N;N;7120;; STR;VN1v2;Dallas, Tx;RTCM 2;;0;;SNIP;USA;32.89;-96.70;1;0;sNTRIP;none;N;N;0;none; STR;ZUPT6818;Houston, Tx;RTCM 3.2;1005(1),1077(1),1087(1),1097(1),1127(1);0;GPS+GLO+GAL+BDS;SNIP;USA;29.94;-95.53;1;0;sNTRIP;none;N;N;0;none;

I pasted the sourcetable from here into a file and did grep. http://www.rtk2go.com:2101/

Nothing seems to show up in PA or OH.

*Thread Reply:* Checking now....trying to figure out the distance. Austin will be West and Houston will be East. (link)

google.com

2126 Happy Hollow Rd

2126 Happy Hollow Rd, Brenham, TX 77833

Original URL: https://goo.gl/maps/5ViSHEvr99u4BhxG7

*Thread Reply:* Looks like 60 miles

*Thread Reply:* Or 73 miles from Austin

*Thread Reply:* It is fascinating that I was just looking at RTK reviews of the ublox devices:

1. https://blog.adafruit.com/2020/06/08/eye-on-npi-u-blox-c099-f9p-application-board-for[…]f9p-gnss-rtk-module-eyeonnpi-adafruit-digikey-digikey-ublox/
2. https://deepsouthrobotics.com/2019/06/03/the-taming-of-the-u-blox-zed-f9p/

Adafruit Industries - Makers, hackers, artists, designers and engineers!

EYE ON NPI – u-blox C099-F9P Application Board for ZED-F9P GNSS RTK Module #EyeOnNPI #Adafruit #DigiKey @digikey @ublox

This weeks’ EYE on NPI is more like EYE to the SKY, as we feature u-blox’s low cost GNSS RTK eval/application boards. Digi-Key has plenty of stock of u-blox ZED-F9P eval boards, which a…

Original URL: https://blog.adafruit.com/2020/06/08/eye-on-npi-u-blox-c099-f9p-application-board-for-zed-f9p-gnss-rtk-module-eyeonnpi-adafruit-digikey-digikey-ublox/

*Thread Reply:* Since I don't know how the position degrades with distance from a base station, I always think it would be good to experiment with that. But I don't have a convenient way to do that. But it just occured to me that I could tell the CORS system that I am by any random base station and it might send me the data from that location. I might try that if I get around to it. I can just dump the binary data to my screen and it has a few readable messages in it. One message is the name/location of the base station. So I could verify that I can actually pick and choose base station sites.

*Thread Reply:* This does seem to work. I can get data from Cologne, MN (32.5 miles) or Hollywood, MN (37.2 miles) or Brownton, MN (57.5 miles). So I can click on the map and it tells me where the base station is. I'll have to figure out how to test this. I'm using this map: http://mncors.dot.state.mn.us/Map/SensorMap.aspx

Oh, if you put in "penn" it shows a couple in Pennsylvania. Are those close to you.

jeff@jeff-hp-pavilion:~$ grep -i penn rtk2gosourcetable.txt STR;Base9557;Is near: Doylestown, Pennsylvania;RTCM 3;PENDING;;;SNIP;USA;0.00;0.00;1;0;sNTRIP;none;N;N;0;none; STR;MAMCCMR1;Is near: Doylestown, Pennsylvania;RTCM 3;PENDING;;;SNIP;USA;0.00;0.00;1;0;sNTRIP;none;N;N;0;none;

I found the path generator code. It is under Juan's files. https://github.com/rje1974/Proyecto-Pochito/blob/master/software/base/tractor/catkin_ws/src/pochito/scripts/path_generator.py I t was referenced in this video: https://www.youtube.com/watch?v=wnXnZdxiRt8 At around 23:50

GitHub

rje1974/Proyecto-Pochito

Software y documentación para automatización de un tractor corta césped. - rje1974/Proyecto-Pochito

Original URL: https://github.com/rje1974/Proyecto-Pochito/blob/master/software/base/tractor/catkin_ws/src/pochito/scripts/path_generator.py

YouTube

} ROS Agriculture (https://www.youtube.com/channel/UCnr6AObsHnHSGPge0HVlBpA)

Lawn Tractor Meeting 07052019

Original URL: https://www.youtube.com/watch?v=wnXnZdxiRt8

*Thread Reply:* My files are a real mess..

*Thread Reply:* I just got "path_planner.py" to run. And it does work.

I loaded it and ran it under python3, so it does work under python3 also.

I copied it into my home directory. I had to load "dubins". pip3 install dubins I had to change the path for the "waypoint.txt" input file. I just made up a random waypoint file. And ran it. I was surprised that it created a graph for me.

It also created a "generated_points.txt" output file.

So, @Al Jones you can modify paths all you want...

*Thread Reply:* This is the output file. jeff@jeff-hp-pavilion:~$ cat generated_points.txt 100.0 0.0 0.0 101.0 0.0 0.0 102.0 0.0 0.0 103.0 0.0 0.0 104.0 0.0 0.0 105.0 0.0 0.0 106.0 0.0 0.0 107.0 0.0 0.0 108.0 0.0 0.0 109.0 0.0 0.0 110.0 0.0 0.0 110.95885107720841 -0.24483487621925448 5.783185307179586 111.78366617911476 -0.8053070937595033 5.836654403443426 112.75423179462344 -0.9985475938345956 0.05346909626383933 113.6986276847929 -0.7028181439855021 0.5534690962638393 114.38563279875153 0.009476372539585998 1.0534690962638393 114.64704432446128 0.963941415955595 1.5534690962638393 114.4188596033827 1.9268906555021608 2.0534690962638393 113.75694621345619 2.662560533309475 2.5534690962638393 112.8233636375723 2.990833405906349 3.0534690962638393 111.84668545013798 2.831336625164389 3.5534690962638393 111.01861614211509 2.2769477785279775 3.6741235767316827 110.06823567732195 2.00105563109248 3.1741235767316827 110.0 2.0 3.14 109.00000000134673 2.000002030529716 3.141593161222221 108.00000000134688 2.000001522897288 3.141593161222221 107.000000001347 2.0000010152648597 3.141593161222221 106.00000000134713 2.000000507632432 3.141593161222221 105.00000000134726 2.000000000000004 3.141593161222221 104.00000000134739 1.999999492367576 3.141593161222221 103.00000000134752 1.999998984735148 3.141593161222221 102.00000000134764 1.99999847710272 3.141593161222221 101.00000000134777 1.999997969470292 3.141593161222221 100.0000000026945 1.999999999995715 3.1400000013472558 jeff@jeff-hp-pavilion:~$

*Thread Reply:* nice, I am a bit lost. When I try to go back to all this, it got complicated .. Today I am going to try again .. I would like to get to Friday with this ready and a small text where I explain what my goals with the robot are. It left me wondering the question you asked about what we thought we were going to achieve and if we were confused with the goals that matt had said were possible

*Thread Reply:* Looking forward to working on it with you guys on Friday.

*Thread Reply:* Al and Jeff, I'm for the meeting, unless it gets complicated with my sister-in-law, who is ill and I have to help my brother.

I think last week the meeting was very good although we got complicated with jitsy .. How about extending the meeting a little and let's work a little with the simulators together ..?

With all this Spanish translations I am with a lot of work and I have not been able to dedicate myself to simulators. I'm also going to start reporting in slack when I'm working in the simulator in case anyone wants to do something together.

*Thread Reply:* My idea is that the robot travels a path, where it stops several times, each time that happens it takes a variable and continues its journey until it ends where it started. With the improvement of these steps, I plan to advance to a robot that does recognition over long distances and on large surfaces autonomously. For those goals I need to bring the following points to the robot that I achieve in the simulator.

1. A simple navigation between two gps points.
2. The value of a variable provided by a sensor when it reaches the selected point.

Those things I achieved in the simulator and if it weren't because my tractor is broken, I could be trying to reproduce them. As I can't solve the tractor break for now. The points that I am looking to work on are those in the worksheet that we shared with Al, to which last week I added making movements by commander akermann.

*Thread Reply:* @JeffS will recognize this as the output from path_generator.py. I have been able to replicate what he did. Yea!

*Thread Reply:* Instructions are in the folder "waypoint publisher"

*Thread Reply:*

*Thread Reply:* @JeffS Running the generated waypoints in the simulator failed. I'm going to try and increase the number of waypoints.

*Thread Reply:*

*Thread Reply:* I adjusted the path starting at 0,0 and only went 10 meters, and it ran the course in the simulator. Will now try a larger square path.

*Thread Reply:* Keep in mind this expects your tracotr to be at the starting location and pointed in the right direction. Also the point spacing may need to match the speed of the vehicle. So 1 meter spacing may be too far for it to keep up.

*Thread Reply:* Yeah the simulator does not end very elegantly so I have to restart it each time; It always starts at that same spot

*Thread Reply:* I don't know what you mean by "simulator does not end very elegantly". I would think it should stop at the end of the path you defined.

But after it stops you can do a 2D Nav Goal to have it drive back to the start. But I don't think the "2D Nav Goal Rotational Tolerance" is set to a reasonable value. So it is not pointing in the correct direction when it stops.

*Thread Reply:* re: not ending elegantly, I can't just run another set of waypoints

*Thread Reply:* This set of waypoints sort of drives a square

*Thread Reply:* I also have the same problem with needing to restart the sim after each time

*Thread Reply:* Doesn't Move Base Flex have a "cancel navigation" function? I don't recall it if is a topic or a service. But I think that would reset the statemachine so you can send it the next path.

*Thread Reply:* I have an instruction file in the Google folder called, "Creating a waypoint plan using path_generator.py". It has a few different path images created by a few different waypoint files. Worth checking out IMO.

*Thread Reply:* I have some thoughts on this, I will get back to you...

*Thread Reply:* This is one way to do a square:

THis is really random..

The other day we were talking about how to see the folders of the files we were looking for .. It is a problem that I had several times .. This is a form .. You search and when you mark it with the mouse you see a bar with the address where it is the file, you click on the directory and it takes you there.

This tells how to display the stage simulator screen when running the simulator: @16:00 https://www.youtube.com/watch?v=-JQ77rXgAoc

YouTube

} ROS Agriculture (https://www.youtube.com/channel/UCnr6AObsHnHSGPge0HVlBpA)

Community Meeting 01072020

Original URL: https://www.youtube.com/watch?v=-JQ77rXgAoc

Sharing my electronics layout in pdf format. I have this in Powerpoint if anyone wants the source document. @JeffS Based on this I am getting power to the RPi from one of the power only ports on the usb hub going into the RPi's, what is it a micro-usb port.

Nice, when you say left-right speed.. It's the wheel enconder..?

*Thread Reply:* Is your question about this?

*Thread Reply:* Yes, these are the sensors to detect speed mounted by the rear wheel axle.

*Thread Reply:* Yeah.. thanks..

On a webinar with these guys

@Al Jones Sorry, but in my drunken stuper I let on that people have moved beyond the ROS Agriculture lawn tractor meetings... But I could not help myself...

*Thread Reply:* I'm happy for anyone to join. I updated the meeting notice. Feel free to forward it.

*Thread Reply:* 💩 Nice...

So if get requests for this, should I forward them on. In particular I am thinking of Jono (Jonathan, and his brother) and Doug and Bob (who are both local to Minneapolis

Sorry, I can barely type at this point...

fyi, catkin_ws uploaded from laptop

So, today, I had a burst of energy. During last Tuesday's HBRobotics Zoom meeting I started thinking about my Lawn Tractor power steering unit The first problem was I could not find my power steering motor. Later I found it in my garage. I gave it random attention. I was thinking I would make a chain drive to run my existing steering wheel shaft. I was going to add a 2:1 or 4:1 reduction between my power steering motor and my steering shaft. After adding a "flag" to the steering motor (a ty-warp) and connecting to a 12v battery, I decided I would be better off with a 1:1 connection. (like everybody else has done) I thought I should just put my steering motor on top, replacing my steering wheel (like Juan did). I found that I has an aluminum tube that will press fit onto the splined shaft on the power steering motor. The outside of the tube is 3/4". So I could either put a sprocket on that, or couple that directly to my steering shaft. I had to remove my steering wheel to determine the shaft diameter. The existing steering shaft is also 3/4". So if I get a pipe that has an inside diameter of 3/4" I can directly connect the steering motor in place of the steering wheel. I don't have a pipe with 3/4" I.D. But I found that 3/4" electrical conduit is close enough. That is what I am using to hold up my "green screen" for my Zoom videos. So I can cut about 3" from that. I may do that tomorrow and see how far I get. If that works then I can have a steering motor attached like Juan has. Then I will have to make a bracket to hold the motor in place. It is supposed to warm up in the next week or two. If it stays above the freezing point I may be able to get something done.

*Thread Reply:* I applaud your efforts. Steering control stumped me forever. Kills me guys like this sell a kit for $750 and you still have to figure out how to connect it.

So if I stop 1 "radius" early (I.e., 10 meters - 1 radius = 8 meters). Radius is defined in the python program. Then I picked a point 1 radius plus a small fudge factor up the next line. That looks pretty close to a single arc. Probably close enough that you lawn tractor won't notice.

Now if you don't care if the vehicle follows a particular vertical path, you just define the horizontal rows. The Dubins planner automatically generates a connecting path through your headland area. But the actual row path is from 0 to 10 and 10 to 0.

To take advantage of that , here is a slightly more elaborate path. This is just defining the end points of the rows. And notice that the varied vertical distance it taken care of by the Dubins planner.

If anyone is interested, the Home Brew Robotics ROS discussion groups starts in one hour:

What: This is an open discussion of all things ROS. When: Tuesday Mar 2 @ 7:00PM PST Where: Zoom https://zoom.us/j/92376743464  Why: Homebrewed Robots!   Beginner or expert, if you're interested in ROS please join us.   Thanks, Camp

I poked at the "one radius" a little more. I thought maybe the heading value in radians needed more precision. That didn't fix it. But if I specify exactly 1 radius short of the corner and around the corner just a hair more than 1 radius distance on the next line. It works. And you can see I added 0.001 to the radius value. The resulting curve looks good and appears to be a single arc.

*Thread Reply:* I'm trying to get a reducing square (i.e. like cutting hay). Still struggling 0.0 0.0 0.0 10.0 0.0 0.0 10.0 10.0 3.14 0.0 10.0 3.14 2.0 2.0 0.785

*Thread Reply:* I'll give it some thought (maybe not a whole lot). Seems like eventually you will want to have a program or a script that will cycle through and generate corner points. (Oh, I forgot to send this...)

*Thread Reply:* So I came up with this: 0.0 0.0 0.0 10.0 0.0 0.0 10.0 10.0 3.14 0.0 10.0 3.14 0 2 0 8 2 0 8 8 3.14 2 8 3.14 2 4 0 6 4 0 6 6 3.14 4 6 3.14 But without controlling the vertical lines I think you won't have enough control. You can see as it gets smaller that the end coverage isn't ideal, And the turn degraded to a "light bulb" turn.

*Thread Reply:* No matter, it is going to get funny as your area tightens up at the end.

*Thread Reply:* Al, do you know what the turning radius of your lawn tractor is? And do you know what turning the radius of your full size tractor will be when running in the hay field? You can change the radius inside of the path_planner.py.

*Thread Reply:* I played with this again. If you are doing a large area it behaves fairly well. (by just defining the horizontal rows) As long as you want your two spacing to match your turning radius. So the row spacing is 2 meters and the turning radius is set to 2 meters. You could set your radius to 3 meters but then you have to space your horizontal rows by 3 meters. Otherwise you would have to define both horizontal and vertical rows and use the more complex "stop short by 1 radius" method. 0.0 0.0 0.0 20.0 0.0 0.0 20.0 20.0 3.14 0 20 3.14 0 2 0 18 2 0 18 18 3.14 2 18 3.14 2 4 0 16 4 0 16 16 3.14 4 16 3.14

@Al Jones, I was watching this video: 04172020 Lawn Tractor Meeting <https://www.youtube.com/watch?v=jsqcQD7W01M> This was one of the many times that you seemed confused about getting the correct TF tree for your physical lawn tractor as opposed to the simulator. Do you want to go through this sometime? I think I have seen this come up enough times that I know how to explain the difference and how to fix it.

Thanks Al for the invite.

*Thread Reply:* welcome and see you at today's meeting https://calendar.google.com/event?action=TEMPLATE&tmeid=XzYwcTMwYzFnNjBvMzBlMWk2MG80YWMxZzYwcmo4Z3BsODhyajJjMWg4NHMzNGg5ZzYwczMwYzFnNjBvMzBjMWc2Z280NGMxbTcxMWoyaDlsOG9xNDhkaGc2NG8zMGMxZzYwbzMwYzFnNjBvMzBjMWc2MG8zMmMxZzYwbzMwYzFnOGwwazJoMWw4bDMzaWRwbTZkMmphaDFrOGtzazRkcGg4Y3IzYWNxNDZoMGpnY3E1NmNvMF8yMDIxMDMwNVQxNzAwMDBaIGp1YW5lZHVhcmRvcml2YUBt&tmsrc=juaneduardoriva%40gmail.com|https://calendar.google.com/event?action=TEMPLATE&tmeid=XzYwcTMwYzFnNjBvMzBlMWk2MG80YW[…]aIGp1YW5lZHVhcmRvcml2YUBt&tmsrc=juaneduardoriva%40gmail.com

*Thread Reply:* https://www.worldtimebuddy.com/?qm=1&lid=5128581,3435910,3117735,5368361&h=3435910&date=2021-3-5&sln=14-15&hf=2

What are the time of the meetings?

*Thread Reply:* Hi Vinny..

*Thread Reply:* That link does not seem to work

*Thread Reply:* Just say it in your timezone, I will convert 🙂

*Thread Reply:* https://www.worldtimebuddy.com/?qm=1&lid=5128581,3435910,3117735,5368361&h=3435910&date=2021-3-5&sln=14-15&hf=2

Worldtime Buddy

World Clock & Time Converter

Effortless time conversion and world time. Schedule conference calls, webinars & online meetings, plan travel and track flight arrival time across time zones.

Original URL: https://www.worldtimebuddy.com/?qm=1&lid=5128581,3435910,3117735,5368361&h=3435910&date=2021-3-5&sln=14-15&hf=2

*Thread Reply:* Thank you Juan

*Thread Reply:* Unfortunately, won't be there this week but have fun all 🙂

Thanks Jeffs..

VUELTAS LOCAS - crazy turns

-14 0 0 0 0 0 40 0 0 40 2.5 3.1415 0 2.5 3.1415 0 5 0 40 5 0 40 7.5 3.1415 0 7.5 3.1415 0 10 0 40 10 0 40 12.5 3.1415 0 12.5 3.1415 0 15 0 40 15 0 40 17.5 3.1415 0 17.5 3.1415 0 20 0 40 20 0 40 22.5 3.1415 0 22.5 3.1415 0 25 0 40 25 0 40 27.5 3.1415 0 27.5 3.1415 0 30 0 40 30 0 40 32.5 3.1415 0 32.5 3.1415 0 35 0 40 35 0 40 37.5 3.1415 0 37.5 3.1415 0 40 0 40 40 0 47 40 4.7123 47 1.25 4.7123 43.5 -2.25 3.1415 -2.5 -2.25 3.1415 -5 0 1.57075 -5 40 1.57075 0 42.5 0 40 42.5 0 42.25 40 4.7123 42.25 1.25 4.7123

*Thread Reply:* Wow. I'm impressed.

*Thread Reply:* Nahhhh, @JeffS explained very well this subject and the rest is practice..

Where is cmdvel linear x coming from? I ran this path, captured the bag file and noticed cmdvel linear vel.x varies from ~1.6 to ~1.9. @vinny ruia I will keep searching, but I'm wondering if you know off hand where "lawntractorsim lawntractorsim.launch" determines the speed to tell the tractor to run if waypoints are sent to it using pathpublisher.py and pathfollowernode.py which use /movebaseflex/exepath? I'm only giving it x,y coordinates in the waypoint planning process.

*Thread Reply:* standby, pulling up the bag now

*Thread Reply:* thanks. We need to basically work through how to program the speeds we want to program the tractor to travel. Knowing where the default is set seems to be a good place to start.

*Thread Reply:* I believe it is being set by the local planner. On the develop branch, you can see that the config file for teb_local_planner_params_carlike.yaml has the max_vel_x set to 2.0

https://github.com/ros-agriculture/lawn_tractor/blob/develop/lawn_tractor_navigation/config/teb_local_planner_params_carlike.yaml

GitHub

ros-agriculture/lawn_tractor

Software for self driving lawn tractor. Contribute to ros-agriculture/lawn_tractor development by creating an account on GitHub.

Original URL: https://github.com/ros-agriculture/lawn_tractor/blob/develop/lawn_tractor_navigation/config/teb_local_planner_params_carlike.yaml

*Thread Reply:* yeah I just pulled up the bag and I am confident that is the relevant parameter

*Thread Reply:* I set the param to 0.75; checking out the bagfile now...

*Thread Reply:* hmmm, does not seem to have influenced cmd_vel. It is much greater (1.88) than the max I set (0.75). 😞

*Thread Reply:* send over the new bag file please?

*Thread Reply:* and, if not too much effort, can you tell me how to recreate? (how are you generating this bag file)

*Thread Reply:* uploading now, same folder as above; using command al@al-ThinkPad-W530:~/bagfiles$ rosbag record -a

*Thread Reply:* sorry, I meant — what are you going to get the robot to drive that path?

*Thread Reply:* @vinny ruia re: recreate, do you want to start from the very beginning and create the waypoint file?

*Thread Reply:* Instructions are here

*Thread Reply:* Ok got these, no I didn't need to create the waypoint file

*Thread Reply:* This is also an important piece to the overall puzzle if you want the full picture. The python source is referenced in that link.

*Thread Reply:* ok, trying to recreate now. Standby

*Thread Reply:* sorry it is taking a bit longer than expected to get sset up

*Thread Reply:* not sure if path_publisher is working correctly,

I keep on getting the message Sending Path when I run it, but when I echo the topic it is publishing on /drive_path, it is just empty poses:

^Croot@pop-os:~/tractor_ws# rostopic echo /drive_path WARNING: no messages received and simulated time is active. Is /clock being published? header: seq: 28 stamp: secs: 0 nsecs: 0 frame_id: "map" poses: []

*Thread Reply:* looks like path_publisher is looking for a message on a topic called got_path .. not sure where that is supposed to be coming from

*Thread Reply:* yeah, good question. Not the author, so a bit of a challenge; Any other basic path publishing routines around you know of?

*Thread Reply:* I am digging into it and trying to delete reference to got_path

*Thread Reply:* I am surprised you got it working without modification .. maybe there is something going on I am missing

*Thread Reply:* ok I got it working

*Thread Reply:* Do you notice the following warnings on the stdout of the window that you called the lawn_tractor_sim lawn_tractor_sim.launch

```[ WARN] [1615069628.314063033, 18.700000000]: TebLocalPlannerROS: trajectory is not feasible. Resetting planner...

*Thread Reply:*

*Thread Reply:* Similar

*Thread Reply:* I wonderr what that is about

*Thread Reply:* If it were an 8 cylinder engine I think we would be firing on about 5 1/2.

*Thread Reply:* Re: Your orignal question: • I was able to recreate and limit the speed by changing the parameter in the config file I mentioned above

*Thread Reply:* Remember that you have to edit the config file and catkin build && source devel/setup.bash in the container or window that calls roslaunch lawn_tractor_sim lawn_tractor_sim.launch

*Thread Reply:* Also, note that the above warning messages disapearred when the I made max_vel_x: 0.0

*Thread Reply:* So you changed /movebaseflex/TebLocalPlannerROS/maxvelx and the vehicle did not exceed the value you set?

*Thread Reply:* let me confirm with a bg ... but the robot was visually alot slower

*Thread Reply:* I used $ rosparam set /movebaseflex/TebLocalPlannerROS/maxvelx 0.75

*Thread Reply:* and then verified with $ rosparam get /movebaseflex/TebLocalPlannerROS/maxvelx

*Thread Reply:* Ah.... I am not sure if that parameter can be changed at runtime

*Thread Reply:* I was not clear how to changed a param like that in Docker

*Thread Reply:* there are a few ways, you can use nano or vim to edit the file in the container... or you can use the vscode docker extention to remote into the container

*Thread Reply:* you might have to CTRL^C out of the launch of that container before you do these edits

*Thread Reply:* and then as always, remember to build your workspace and source your setups

*Thread Reply:* When I rerun the bag file the param certainly looks like it "took".

*Thread Reply:* I believe you that the paramater "takes", I just have the funny feeling that the underlying planner will not "listen" to parameter changes

*Thread Reply:* here is a rosbag that is proof that the maxvelx set to 0.5 worked

*Thread Reply:* in the lawn_tractor container, the full path to the parent dir is /root/tractor_ws/src/lawn_tractor_navigation/config

*Thread Reply:* See if I have the same as you /root/tractor_ws/src/lawn_tractor/lawn_tractor_navigation/config

*Thread Reply:* yep, looks the same

*Thread Reply:* here is my yaml ```# file: teblocalplanner_params.yaml TebLocalPlannerROS:

# Trajectory tebautosize: True dtref: 0.3 dthysteresis: 0.1 globalplanoverwriteorientation: True allowinitwithbackwardsmotion: False maxglobalplanlookaheaddist: 3.0 feasibilitycheckno_poses: 5

# Robot maxvelx: 0.50 minvelx: 0.0 maxvelxbackwards: 1.0 maxveltheta: 2.0 acclimx: 2.5 acclimtheta: 2.5 minturningradius: 1.0 # diff-drive robot (can turn in place wheelbase: 0.4 cmdangleinsteadrotvel: True weightkinematicsturningradius: 1.5 # increase of minturningradius is not enough footprintmodel: type: "line" # include robot radius in minobstacledist linestart: [0.0, 0.0] # include robot expanse in minobsticaldist lineend: [0.4, 0.0]

# Goal Tolerance xygoaltolerance: 0.4 yawgoaltolerance: 0.2

# Obstacles minobstacledist: 0.25 costmapobstaclesbehindrobotdist: 1.0 obstacleposesaffected: 10

# Optimization noinneriterations: 5 noouteriterations: 4 optimizationactivate: True optimizationverbose: False```

*Thread Reply:* I agree that it is confusing that setting the rosparam didn't work, but I am sure that setting that parameter at run time is not something the developers imagined people would ever do

*Thread Reply:* so I bet they probably just take the value of that parameter when the planner first initializes, and sticks with it.

*Thread Reply:* I am going to have to sign off soon, any other doubts I can help with?

*Thread Reply:* bash: nano: command not found

*Thread Reply:* sudo apt-get update && sudo apt-get install nano

*Thread Reply:* vim is installed

*Thread Reply:* if you are comfortable with vim go for it 🙂

*Thread Reply:* I just wasn't sure of your experience

*Thread Reply:* I'm good; thanks for your help; Have a great evening

*Thread Reply:* You too :)

*Thread Reply:* btw, what did you change in path-publisher.py?

*Thread Reply:* @Al Jones I am sorry I just noticed this

*Thread Reply:* I don't think anything was changed in path_publisher

*Thread Reply:* oh sorry, yes I did

*Thread Reply:* https://github.com/ros-agriculture/lawn_tractor/tree/feature/add-path-publisher

GitHub

ros-agriculture/lawn_tractor

Software for self driving lawn tractor. Contribute to ros-agriculture/lawn_tractor development by creating an account on GitHub.

Original URL: https://github.com/ros-agriculture/lawn_tractor/tree/feature/add-path-publisher

*Thread Reply:* the latest is on that branch, I am PR-ing it with Matt as reviewer... but not sure if he is available.

In that case who shall I assign to review?

*Thread Reply:* @vinny ruia Can you assign 3 names with "or's"? Meaning any of @JeffS, @Juan Eduardo Riva or @Al Jones could approve?

*Thread Reply:* i gave it to @Juan Eduardo Riva but next time I wiull do that

I have my first group of meetings indexed. I have 100 of 250 done. At various times I will just overwrite the file as I get more done. This is just a temporary location. Eventually I will find a permanent home for it. I may try to figure out how to get the actual links for the "Chat:" sections. I may figure out a way to convert to a different format so the links are actually clickable. I will put a README file in the same directory explaining what I did and how to use it. Eventually that type of stuff should be merged into the index file. This seems to be accessible with Ubuntu 20.04/Firefox and on Windows 10/Google Chrome. sampson-jeff.com/RosAgriculture

It turns out if I create a file named README or README.txt that the browser does not see it. But if I call it readme.txt then it shows up in the browser. So there is a readme.txt file in the same directory.

I experimented by copying your long text file in to my iPad. Then I pasted that into my iPad’s word processor called PAGES. So far so good. Next I tried high lighting a couple of the http addresses and made them links. It worked. Reading thru your document, I’ll be able to click on the new http link and go right to the YouTube meeting. Guess I know what I’ll be doing in my free time. Thanks Jeff for making it possible.

If you are modifying the file, keep in mind I will totally replace it when I do updates. So give some thought as to how much work you plan to put into that.

I think I’ll wait. Thanks for the heads up.

This message about Docker showed up on the HBRobotics group. I don't know anything about Docker and I don't plan to learn it.... But this may mean something to people that use Docker.

In this meeting, during this Docker discussion the issue was raised as the what happens if changes are made in a Docker Containing while it is running, such a revising script in the ROS workspace src or sudo update, then shutting down and restarting. Are the changes persistent? In answer to that, Steve referred to a dos.docker.com/bind-mounts document that described using the -v flag, which the docker run in this case uses. Following up with the developer, Raffaello sent me the following reference that seems to further address the question. "You need to terminal from the nanosaur host to store the changes", https://phoenixnap.com/kb/how-to-commit-changes-to-docker-image. "If you want share part of the code between volumes, you need to run the Container with a volume, in this case you will able to move files inside and outside the container". I haven't tried this but its on my ToDo list. Ross

Knowledge Base by phoenixNAP

How to Commit Changes to a Docker Image (With Example)

In this Docker Tutorial Learn how to commit changes to a Docker image. Save changes to a running Docker container and create new images now with examples.

Written by

Sofija Simic

Est. reading time

3 minutes

Original URL: https://phoenixnap.com/kb/how-to-commit-changes-to-docker-image

*Thread Reply:* yep, it's a pretty clear document on how to deal with Docker. I'm going to translate it into Spanish to have it handy for the ROS wiki tutorials. Thank you very much

In the last few days I was indexing an old meeting video and we were talking about Al's absolute sensor which is single turn. Yet it has to turn 1-1/2 to 2 turns in normal operation. It was suggested to add an additional multi-turn pot to determine what the overall absolute position is. I thought of another way that may work. Use those magnetic sensors like Matt is using on his back wheels. Strategically mount them above the "half moon" gear. If it is centered then neither sensor is on. If turned some amount to the right them the right senosr will turn on. If turned some amount to the left then the left sensor will turn on. If neither sensor is on then you know you are in the center section. But I can't visualize from memory if you would be able to mount them properly.

I: 06122020 Lawn Tractor Meeting https://www.youtube.com/watch?v=OED92oZw9ts At the very begining of the video is where we are talking about this.

YouTube

} ROS Agriculture (https://www.youtube.com/channel/UCnr6AObsHnHSGPge0HVlBpA)

Lawn Tractor Meeting 06122020

Original URL: https://www.youtube.com/watch?v=OED92oZw9ts

*Thread Reply:* @JeffS Let me pull up the video of the motor mount

*Thread Reply:* For me, the problem statement is I need to know the wheels are pointing straight before engaging the magnetic sensor since the magnetic sensor does not retain its position data on reboot. The solution has been a "low resolution" analog encoder and some code to test if the angle is within the "straight zone". btw, I'm still not sure having a "high resolution" magnetic sensor is worth the effort yet. Below is a photo clipped from the video.

*Thread Reply:* @JeffS Came across this...https://youtu.be/JyIkXSQGX_8?t=162

YouTube

} Michael Rechtin (https://www.youtube.com/channel/UCHS55yDvORmCpbM_3vNQQsQ)

Homemade DIY Remote Controlled Lawn Mower!

Original URL: https://youtu.be/JyIkXSQGX_8?t=162

*Thread Reply:* I did see this this, even though it was buried in a thread, since you put my name on it. I see part two is talking about the electronics and remote control.

*Thread Reply:* @JeffS Yes, "@'ing" people does help.

I though of a new concept. "Words of Wisdom"...

Words of wisdom: "Try some something quick and cheap to verify that is what you want to do". Words of wisdom: "You are going to need a bigger box than you think".

*Thread Reply:* "You are going to need a bigger box than you think", that happened to me becouse I'm cheap..

and..

"Try some something quick and cheap to verify that is what you want to do". THis is a MOD in life.. but this one has a sister.. I don't know how to put this in a way that is well understood; But we are going to try: "accept when you buy I buy bad things because they are miserable and not waste time trying to resuscitate them", also here in Argentina it happens that we buy "what there is" and not "what we should" so working with things that are not optimal..

I was watching another video and I was remined that we should go over A/D conversion sometime. Which reminds me I need to find an op-amp that will run on either 3.3V or 5V and correctly give the correct output.

I was looking at the latest zoom meeting. Some one was looking for ideas to lift up their heavy lawn tractor. Maybe use a modified sheep grooming stand ...like this crank up one or hydraulic lift. Just narrow the area that the sheep stands on. Match it to the underside of the lawn tractor..... https://www.valleyvet.com/ctdetail.html?pgguid=88bbbde3-9bde-47fc-8850-9e73e487870e&itemguid=a17c98ac-9374-4513-bd0e-892521a97888&sfb=1&grp=L000&grpc=L600&grpsc=L640&sp=f&utmcontent=35673&ccd=IFF003&gclid=CjwKCAiA4rGCBhAQEiwAelVti7sz-FdyRvEtaUR1cLIk2rsX1coh5I6V4-GqFaglTOR3z4DjQdjyBoCkS8QAvDBw|https://www.valleyvet.com/ctdetail.html?pgguid=88bbbde3-9bde-47fc-8850-9e73e487870e&itemguid=a17c98ac-9374-4513-bd0e-892521a97888&sfb=1&grp=L000&grpc=L600&grpsc=L640&sp=f&utmcontent=35673&ccd=IFF003&gclid=CjwKCAiA4rGCBhAQEiwAelVti7sz-FdyRvEtaUR1cLIk2rsX1coh5I6V4-GqFaglTOR3z4DjQdjyBoCkS8QAvDBw

*Thread Reply:* Thanks.. I'm looking this two options..

🦾

I copied my catkin_ws and a few files from my root directory on my RPi to the Google drive.

I'll go look at it later. It is time for the Home Brew Robotics Group meeting. Or I may look at it while that meetig is going on...

Okay, so I am looking at Al's RPi catkin_ws. I can see files. If I click on one it goes down that path of "what do you want to open it with... I say text editor... It wants me to agree to endless questions if I want to sign away my life..." About 3 steps into I gave up. I can download the file. Then force it to open. Is there an quick way to view a file?

*Thread Reply:* @JeffS If I right click .bashrc and open with TextEditor it just opens. I'm not sure what you are being presented with.

Just the usual making ROS2 work.

Hi Jeff, here is the tool to mount the google drive directories on your computer and use the editor you want https://support.google.com/a/answer/7491144?hl=en#zippy=%2Cwindows

Is there a link for today’s lawn tractor meeting?

*Thread Reply:* I was about five minutes late for the meeting and some of Jeff's news may be missing; but if I did not understand something wrong or I forget anything, things were this way ...

Jeff showed the model that he is assembling in conglomerate wood on which he is going to mount the steering of the vehicle and the other accessories and electronics, as if to start his tests. I found it very interesting and I'm sure I'll copy it shortly. He explained to us how he is going to compose the entire steering column and he showed us his gear where he is going to connect the steering motor and the potentiometer. We were talking a little about the files that are in google drive and how to use them better, to which google drive opens a program to see the code of the python files. I told them that if you mount the directories in the root with the specific app for w10 you can use vscode. Then Al showed us his new 3D impressions for the wheel magnetometer sensor from which he receives odometry. It was very good and super neat. It will be necessary to see how long these gears last with a little dirt and the other elements but clearly enough to test in the field. The meeting lasted about 25 minutes, because Al was about to catch a plane. I said that: a boy began to help me, take me to fix the tractor (they give it to me on Monday) and that I began to present the electronics again with my assistant.

*Thread Reply:* @Bob Hassett To get @JeffS’s attention you want to put "@" in front of his name.

*Thread Reply:* And I see I can't reply to a reply in a thread... Bob, no I can't (conveniently) post a pic of what Juan was talking about. I did show it at both the ROS Agriculture meeting Friday morning and at the TCRobots meeting the night before. Both of which you belong to. As lazy as I am I will probably not get the motor mounted before next Friday's meeting, so I may show it again there.

*Thread Reply:* ok if we don't like the thread idea we can scrap it... I won't be offended 🙂

*Thread Reply:* I'm not saying it should not be done. I am just saying that Bob has founsda way around it.

No, we could not figure out a reasonable way to record it and it lasted 30 minutes.

Can someone provide a link to today’s meeting? Will it be recorded?

https://us02web.zoom.us/j/82088036016?pwd=K2lLc1FiWm9MU0dzRStxM2J2b3dpQT09 Bob, we will try this one, if it does not work we will use other one and i will send it here.. It's ok..?

Can not make it work.. Please use this link https://us02web.zoom.us/j/84606675957?pwd=S25tRVpsb2tZUnQ1SGFXdEdPbjkzZz09

*Thread Reply:* Not much.. Hope you are ok.. Bye..

This is the page that inspired me to make my steering motor frame for my lawn tractor. https://www.sparkfun.com/news/3329

Young victim..

I watched a video about the Bosch BMO055 IMU module: https://www.youtube.com/watch?v=2AO_Gmh5K3Q I was inspired. I had two sensors that were cheap crap that I bought on eBay or Amazon. I wired one up and could not figure out how to get the Adafruit libraries to work. So I tried this library: https://github.com/kriswiner/BNO055 I spent hours trying to get anything out of it. After switching to a second board and soldering two jumper pads it seems to be working. I think the first board has a short. the regulator got hot enough to burn my finger... Once it was running the yaw appeared to be stable. But it did not come up the same orientation each time I started it. And this particular code had me wave the board around like an idiot. It seems like it would be hard to convince your robot to wave it around when it starts up. Maybe I can figure out how to store the calibration values so it does not need to be done each time.

But it is running. I need to read about it and see what can be done to make it usable.

YouTube

} Paul McWhorter (https://www.youtube.com/user/mcwhorpj)

9-Axis IMU LESSON 1: Introduction to Nine Axis Sensors and Inertial Measurement Units with Arduino

Original URL: https://www.youtube.com/watch?v=2AO_Gmh5K3Q

GitHub

kriswiner/BNO055

Teensiduino sketch for 9-axis BNO055 motion sensor + MS5637 pressure sensor - kriswiner/BNO055

Original URL: https://github.com/kriswiner/BNO055

I got a "Tindie" IMU board to run on a Teensy 3.2. It took forever... https://www.tindie.com/products/onehorse/ultimate-sensor-fusion-solution-lsm6dsm-lis2md/ Now I have to decide which IMU to waste my time on. I need to figure out one of the two so I can get one that runs correctly.

Tindie

Ultimate Sensor Fusion Solution - LSM6DSM + LIS2MD by Pesky Products on Tindie

EM7180 Sensor Hub coupled with the LSM6DSM + LIS2MDL IMUs provides < 2 degree heading accuracy!

Original URL: https://www.tindie.com/products/onehorse/ultimate-sensor-fusion-solution-lsm6dsm-lis2md/

Started working on tractor 2 in Texas. Starting with the steering motor mount.

A link for the meeting?

*Thread Reply:* Join Zoom Meeting: https://us02web.zoom.us/j/82088036016?pwd=K2lLc1FiWm9MU0dzRStxM2J2b3dpQT09

$15K https://directedmachines.com/index.html

This is the write-up I follwed to get the "Tindie" IMU to work on a Raspberry Pi: https://github.com/droter/em7180_imu

GitHub

droter/em7180_imu

ROS IMU driver package for the Ultimate Sensor Fusion Solution - LSM6DSM + LIS2MD MPU9250 EM7180 Sensor Hub IMU - droter/em7180_imu

Original URL: https://github.com/droter/em7180_imu

Here is a random series on Cub Cadet mowers. https://www.youtube.com/watch?v=KopYghpcjIk

YouTube

} HAXMAN (https://www.youtube.com/c/HAXMAN)

Was I WRONG About CUB CADET? Watch Before Buying!

Original URL: https://www.youtube.com/watch?v=KopYghpcjIk

Here is a video about sensors on cell phones: https://www.youtube.com/watch?v=C7JQ7Rpwn2k A couple more that may be related: https://www.youtube.com/watch?v=nOQxq2YpEjQ https://www.youtube.com/watch?v=OPsVr44uCb8

YouTube

} Google TechTalks (https://www.youtube.com/user/GoogleTechTalks)

Sensor Fusion on Android Devices: A Revolution in Motion Processing

Original URL: https://www.youtube.com/watch?v=C7JQ7Rpwn2k

YouTube

} WorkLearnShareDevelop (https://www.youtube.com/channel/UCDm91EH9sO_t1Xiuwtlciqw)

How-to create a Compass App

Original URL: https://www.youtube.com/watch?v=nOQxq2YpEjQ

YouTube

} Richvale Consulting (https://www.youtube.com/channel/UCsfKM52q6KXOoHh9FowborQ)

Programming Android Motion Sensors

Original URL: https://www.youtube.com/watch?v=OPsVr44uCb8

Thanks for all the links.

nice links.. learning more American accents.. 🦾

Guys I need to be with my son at his school at Noon so will be unable to be on the call today. I am hoping to get my tractor out this weekend (Pittsburgh) and test a fully planned mission. I was in Texas last week and worked on installing the steering motor and transmission control servo on Tractor II. Something new I did was wire an LED and reset button to the Nano that I install inside the SuperServo for the transmission.

*Thread Reply:* Al, the super servo is for the steering wheel..?

*Thread Reply:* Transmission control

I have not "looked under the hood", pun intended, but thought I would share... https://github.com/udacity/self-driving-car/tree/master/steering-models/steering-node

GitHub

udacity/self-driving-car

The Udacity open source self-driving car project. Contribute to udacity/self-driving-car development by creating an account on GitHub.

Original URL: https://github.com/udacity/self-driving-car/tree/master/steering-models/steering-node

In the real world where should the Teb Planner Run? I'm trying to run a simple path. I'm thinking I need to update my RPI on the tractor with the latest version that matches the code running in the Docker container and continue to use "export ROSMASTERURI=http://tractor_aj:11311" or set "export ROSMASTERURI=http://localhost:11311" and have the RPI just be a slave. Any thoughts?

*Thread Reply:* Not good. Updated the RPI bashrc with ROSMASTERURI=http://localhost:11311. The result is a "rostopic list" command does not return the topics running in the Docker container. I can ping the laptop (aka controller_aj) from the RPI.

*Thread Reply:* $rostopic list on the physical laptop returns all the topics running in Docker.

*Thread Reply:* reviewing: https://answers.ros.org/question/298602/how-to-have-topics-publish-and-subscribe-both-ways-between-host-and-docker-container/

*Thread Reply:* Seems the Docker container/simulator is not exposing an IP Address:

*Thread Reply:* I don't use Docker so I can't help you...

*Thread Reply:* @JeffS Do you launch the "develop" branch of lawntractor? How do you do that in a roslaunch lawntractorsim lawntractorsimreal_world.launch command? The reason I ask is because I think there are changes in develop that are needed that are not in master.

*Thread Reply:* @Al Jones I don't run anything. I don't know what you are asking here: "How do you do that in a roslaunch lawntractorsim lawntractorsimrealworld.launch command?" I can see changes on this page: https://github.com/ros-agriculture/lawn_tractor/compare/develop What are the errors you are getting? Vinny's name is on the "develop" branch. He probably knows how this works.

GitHub

ros-agriculture/lawn_tractor

Software for self driving lawn tractor. Contribute to ros-agriculture/lawn_tractor development by creating an account on GitHub.

Original URL: https://github.com/ros-agriculture/lawn_tractor/compare/develop

*Thread Reply:* @JeffS Thanks. Re: error messages, I'm not getting any. My problem statement at the moment is the RPi (192.168.193.95) is not seeing the topics published by the ROSMASTER which is the instance of ROS running in the Docker container on my Ubuntu laptop (192.168.193.96). I think my next step will be to add the IP address of the laptop to the ROSMASTER statement on the RPI. If that does not work I will try and run the simple mission on my native laptop without Docker. I have forgotten why that did not work the last time.

*Thread Reply:* On Laptop: export ROSMASTERURI=http://localhost:11311; on RPi: export ROSMASTERURI=http://controller_aj:11311 seems to do the trick.

I updated the instructions a bit on using the waypoint publisher. Mostly adding samples and reference to the video where Jeff spoke about creating his model.

Had a minor setback last night. I spent so much time making sure my control box was wired correctly but I for got to set my go no go toggle properly. I blew one channel of my motor controller. so I guess I need to order a new board.

DPDT on off on was in the wrong position. It fed 25 volts back into the controller when I pressed the test button on the board.

I don't see a problem. Your schematic isn't quite complete, but I get the idea. I don't see any combination that could blow anything out. I think you have a wiring error. Which ever side blew out, I think the switch it is connected to is wired wrong.

I’ll go thru the wiring again and use a test light as I go. Thanks for the feed back.

Every thing working as it should be. Switches where ok, so I must have mixed up the primary power lines some how. Reminded to always use a test light first and assume NOTHING. Still need to replace the driver.

I’m all for robotics in agriculture BUT it @should not be used to promote gigantic destructive monoculture farming systems. Robotics should be used to create “a revolution in farming practices”. Just saying. https://www.geekwire.com/2021/isilon-founder-lifts-hood-farming-startup-carbon-robotics-weed-zapping-machine/?fbclid=IwAR3wT7-fPhT9pPhPYbATmXiS4U8LgCpZuxsVtNht7HFYECJLpHEsDa7ctf4

GeekWire

Isilon founder lifts the hood on farming startup Carbon Robotics and its weed zapping machine

Carbon Robotics, a Seattle company led by Isilon Systems co-founder Paul Mikesell, is unveiling its self-driving robot that uses artificial intelligence to identify weeds growing in fields of…

Original URL: https://www.geekwire.com/2021/isilon-founder-lifts-hood-farming-startup-carbon-robotics-weed-zapping-machine/?fbclid=IwAR3wT7-fPhT9pPhPYbATmXiS4U8LgCpZuxsVtNht7HFYECJLpHEsDa7ctf4

*Thread Reply:* I can be wrong but what I see that the robots are going to generate is a virtuous consociation of crops that is otherwise impossible. Today agriculture already has very modern techniques such as the integrated management of insects that with robots can be taken to another level. All these techniques are of processes and for the taking of decisions a lot of information that is gathered in the field is made flat, unfortunately the path that was chosen for development was wrong at first; but fortunately that is being rectified

*Thread Reply:* I think we have to wait a little longer, the issue goes through the artificial intelligence that is being developed .. You have to look at that .. the robots will only be the means for the execution of that artificial intelligence. Look at the new data sets, none of them are from the big companies and most of them are free. For example, the only case of a useful robot in the agricultural sector of Argentina is one that does plant physiology.

Is there a meeting today? A link?

Yes, link is at the top of the General channel and I will email you an invite.

I assume there may be a meeting today. Unless people start canceling. As always, the link is still in the pinned message at the top of the General channel, same place it was the last two times you asked.

I need your email address

Who's email address?

Oh, I see, you were asking Bob (due to multiple messages Slack confused me again).

Everyone should consider that you can send a private message to someone. In case you don't want your email address (or anything else) to be public. When Al sends you the invite you can go back and delete your message if you don't want it to be public...

So to understand the link in General? I just click on the link and wait?

Yes

Ok I’ll do that.

This is my first control box from last year. Biggest problem was I used heavy primary wire so it was difficult to close up the box.

Using the control box to drive backwards while operating a snowblower on the 3 pt hitch.

This year’s control box. Black momentary switch from left to right...steering, forward/reverse, throttle, 3 pt hitch.

Here is an article on Farmhand robot: https://www.sparkfun.com/news/3780

So the spark fun extended range they are talking about is up to 100 ft of wire?

Why do you ask? What do you think they are trying to do?

The term “extended range”? I’m not sure what they are trying to do? Just use a distributed system for a mobile machine? So they are using Ethernet wire on a single “tractor” ? Instead of using usb and a hub?

It all comes down to the fact they are trying to inappropriately use a communication bus for the wrong thing. It goes back to the fact that I2C was designed as a bus between two or more chips on a single printed circuit board. But it is real handy to get a sensor that uses I2C and connect it to a microcontroller over a length of wire. The I2C bus drive the bus low and pulls to an inactive state using pull up resistors. Which works fine if the chips are close together on a printed circuit board. But when you separate them with wire, you re adding inductance and capacitance which will affect the signal propagation. If the wire gets too long the signal degrades and you will start getting random errors. If the wire gets too long it won't work at all. If you try this (connecting I2c devices on medium to long wires) on a vehicle (or any other electrically noisy environment) then you are introducing electrical noise to you bus which it was never designed to handle.

To further complicate matters. The I2C has two wires. A clock line and a data line. I think the clock is always driven by the master. The data line is bidirectional. So if you want to add line drivers/receivers to the existing signals you have to account for which device is in control. And I2C used some clever (foolish) tricks to to make it work (basically "clock stretching" was added as an afterthought for accommodate newer slower parts. And they watch the state of the bus to decide it they are allowed to transmit or not. And to decide if the bus is hung So Basically, none of this was designed to run over an extended length of wire.

So I didn't realize until I saw this article that someone (Philips?) designed a chip to get around (some/maybe all) of these problems. The PCA9615 is designed to be I2C on one side and has differential driver/receivers on the other side. It apparently has all kinds of magic designed inside the chip to make it all work. So they have provide a method to get an I2C signal to work over an arbitrary distance.

To relate a similar story, at Cray they were building Super Computers. Not the old vector processors, but new "lots of CPU core in a box" concept. One on the latest cabintes I was associated with (from memory) had 96 nodes. Each node was a dual processor SMP (symmetrical multiple processing) configuration using AMD Athalon server CPUs. So that meant it had 192 AMD server class CPUs in a single cabinet. Each node was basically a full dual CPU desktop or server computer, minus about half of the circuitry because you don't need keyboard/monitor/disk/USB/networking. But they had their own big networking for each node. It had big power supplies, redundant power supplies. I think each cabinet drew about 54KW when it was thinking hard. And the cooling fan took 5KW to keep it from melting down. Well, with big 3-phase cooling fans and huge power supplies it was relatively electrically noisy. So someone decided to put a microntroller in the bottom of the cabinet. Then run a cable to the top of the cabinet to connect to a random I2C sensor (I think it was a temperature sensor). Guess what? It didn't work well. So here was real world example of a solid brick about the size of a refrigerator drawing 50KW and it had some of the same problems you will have on a lawn tractor...

Bottom line, I don't use I2C unless I have to, A reason I would have to use it is if I buy a sensor that has that interface. But if have to use it I would put the sensor and the microcontroller as close together as possible and only add a length of wire if I need to. An example of that would be if I have a magatometer that I don't want close to anything which would affect the magent field. If that becomes a problem then you could put a microcontroller next to each sensor and connect everything together with USB, which is what Al has done.

But basically, I2C started as a cute/simple/elegant solution to put multiple slave peripherals on a single two-wire bus on a localized circuit board. Then it grew wildly ouf of control. Not people are trying to bastardize it even further by trying to create a long range general purpose communication bus...

To answer your other question "So they are using Ethernet wire on a single “tractor” ? Instead of using usb and a hub?" Yes, Sparkfun gave them QwicBus (a big hammer) so everything else looks like a big nail. So rather than switching to a more appropriate method to connect distant nodes together the are going to extend their QwicBus and live it. In my opinion they are trying to force something they shouldn't. Although I now know about the PCA9615 chips that will allow more reasonable solutions in real world solutions. I will have to see if I can find that chip on a simple/cheap breakout board. Because in some cases it would be useful.

[I'm tired of proof reading, here you go.]

If anyone thinks about going down this path, keep in mind the "differential" board (BOB-14589) is different than the "endpoint" board (COM-16988). So make sure you understand the difference before you order anything. https://www.sparkfun.com/search/results?term=pca9615

Not for me. I’ll go with the usb connections. Not knowing anything I really appreciate the explanation! Thanks for taking the time to explain stuff Jeff!

Getting MoveBaseFlex to run on RPi... with tremendous help from @vinny ruia I have cloned the development branch of LawnTractor to my RPi and ran catkin_make cleanly. I have ver. 0 of an RPi launch file (attached). I am getting this error, but will continue to try and figure this out. The rough steps I followed are here.

*Thread Reply:* sudo apt-get install ros-kinetic-mbf-costmap-nav

*Thread Reply:* sudo apt get install ros-kinetic-map-server

*Thread Reply:* try both those installs, and try again

*Thread Reply:*

*Thread Reply:* yeah this would have been faster if we had gotten rosdep update to work

*Thread Reply:* sudo apt-get install ros-kinetic-teb-local-planner

*Thread Reply:* sudo apt-get install ros-kinetic-mbf-costmap-core

*Thread Reply:* sudo apt-get install ros-kinetic-global-planner

*Thread Reply:* remember the patterns I described when you get package failures like that , it’s always ros-<distro>-pkg

*Thread Reply:* I can run rosdep update now

*Thread Reply:* should I do that?

*Thread Reply:* yes. that might be easier than manually installing like we are

*Thread Reply:* rosdep install --from-paths src --ignore-src

*Thread Reply:* this is done at the root of the workspace, after a “rosdep update”

*Thread Reply:*

*Thread Reply:* ok now try the rosdep install command from a few messages up

*Thread Reply:* this one: rosdep install --from-paths src --ignore-src

*Thread Reply:* yup

*Thread Reply:* rosdep update updates the internal sources list , and install goes and installs the dependencies that you want (takes it from the package.xml in your workspace )

*Thread Reply:* bucket load of updates....still running...

*Thread Reply:*

*Thread Reply:* looking better...

*Thread Reply:* current problem statement: I issue a 2d nav goal in RVIS, but I do not get any output on cmd_vel and the State Machine says it is still waiting for a Goal. ...for another day

*Thread Reply:*

*Thread Reply:* Oh for some reason you have the old state machine getting launched

*Thread Reply:* <node pkg="lawn_tractor_navigation" name="mbf_behavior_tree" type="movebaseflex_bt_node" respawn="false" output="screen" />

*Thread Reply:* You can reference this (this is for SIM)

https://github.com/ros-agriculture/lawntractor/blob/develop/lawntractorsim/launch/lawntractor_sim.launch

GitHub

ros-agriculture/lawn_tractor

Software for self driving lawn tractor. Contribute to ros-agriculture/lawn_tractor development by creating an account on GitHub.

Original URL: https://github.com/ros-agriculture/lawn_tractor/blob/develop/lawn_tractor_sim/launch/lawn_tractor_sim.launch

*Thread Reply:* So you are saying line 29 in my launch file is incorrect? <node pkg="lawntractornavigation" name="mbfstatemachine" type="mbfstatemachine.py" respawn="false" output="screen"/>

*Thread Reply:* Yeah. I am not sure if that will fix the problem., but it is a good place to start.

*Thread Reply:* This is better?

<node pkg="lawntractornavigation" name="mbfbehaviortree" type="movebaseflexbtnode" respawn="false" output="screen" />

*Thread Reply:* Yeah try that

*Thread Reply:* roger that

*Thread Reply:* What about this? Need it?

<!-- ******** Global Parameters ********** --> <param name="/usesimtime" value="true"/>

*Thread Reply:* No, do not set that parameter either

*Thread Reply:* Now this looks like serious progress.....

*Thread Reply:* something on cmd_vel and a path planned

*Thread Reply:* @vinny ruia Thank you again...no way would I have made any of this progress without your help; My promise is to write it down and share it with others....

Trying to startup an old raspberry image...buster but when asked for a password the default...raspberry...doesn’t work. I’ve tried holding down the sift key as well as trying shift alt delete. Looked on google and the pi forums but no help. I suppose I’ll have to reflash it? I would lose two other programs on it...Python and openCV

*Thread Reply:* Can you hook it to a monitor..?

*Thread Reply:* Yes, you use Linux commands in the terminal window.

*Thread Reply:* @Bob Hassett Another option, if the RPi is on your network is to use FileZilla from another computer to connect to the RPi if you want to review the files that are on the RPi.

Hello and welcome!

Just for clarity using ubiquity.....I disable their machine.....now I can start going thru the ros tutorials at ros.org? Create a workspace for learning? Create a different workspace for my tractor?

I don't know. What do the tutorials tell you to do? I would just start at the beginning and continue until something conflicts.

Thanks

Just purchased a Lenovo Think Pad X1 Carbon for a robot development work station. It’ll be here in a couple of days. Comes with Windows 10. I’ll be using a raspberry pi in a similar capacity as Lawn Tractor projects. Soooo I’m trying to decide how to set it up.

Partition to go between Windows and Ubuntu?

Dump windows and just do Ubuntu?

Put either of them on a start up thumb drive?

Pros and cons?

The best is to try to get the dual boot working. Both systems is the best and my favorite conf is Ubuntu and W10 + Ubuntu virtualization..

My Lenovo Think Pad X1 Carbon came this afternoon, a day early. Got it connected to our WiFi and the net using the default Bling. Seems I have 5 G of free ram and 214 G free of disk space. So I’m considering Juan and Jeff’s suggestions for downloading as discussed at the meeting today. Is there a link yet of today’s recording?

Found the recording.

First Attempt at 2D Nav Goal....had to kill the tractor, cmd_vel is still sending a reverse signal....I have a bagfile, but will test in a more open space

*Thread Reply:* I moved to a bit more open space, gave the tractor a simple 2D nav goal, but it was not able to navigate straight, forward about 5 meters. After a few minutes I stopped it. It would move back and forth and in a circle, but not straight to the goal. Bagfile is here.

*Thread Reply:* • rostopic echo /movebaseflex/getpath/goal • rostopic echo /movebasesimple/goal • rostopic echo /movebaseflex/currentgoal All return the same goal of pose: position: x: 21.1926879883 y: -1.35362887383 z: 0.0 orientation: x: 0.0 y: 0.0 z: 0.0124892463647 w: 0.999922006321

*Thread Reply:* Congratulations @Al Jones. Maybe you want to try in a near future to only let the tractor move forward from the teb planner as a test..

Congratulations! You have advanced to the point where you are dangerous! 🙂

*Thread Reply:* You are right and here is where all the talking about security begins to have a meaning. As we all move forward I hope that we talk a lot this issue this 2021, but mainly I think that the best approach is not to build something that you can't kick and stop it, it's faster than a one year old kid and weight more than 25 kilos.. 🤷‍♂️:skintone4:💩

*Thread Reply:* For security i¿m starting the navigation test in gazebo first, and next with a small robot with the same hardware (electronics).

I just stumbled on something interesting. I was just playing with my power steering motor and the Cytron 30A power driver board. I realized the trim pot on the board controls BOTH the speed when you push the "jog" pushbuttons and also the speed when you apply an external PWM signal. I'll see if I can track down some documentation on it.

Here is some documentation: https://www.amazon.com/Cytron-30A-Motor-Driver-MD30C/dp/B07L6HGFWY https://www.cytron.io/p-30amp-5v-30v-dc-motor-driver A manual here: https://docs.google.com/document/d/178uDa3dmoG0ZX859rWUOS2Xyafkd8hSsSET5-ZLXMYQ/view#

I still don't see anything saying the pot also controls the external PWM input, but it does. I'll continue playing with it... But this is a really fabulous and affordable motor driver board.

Cytron Technologies

30Amp 5V-30V DC Motor Driver

Note: MD30C is Revision 2.0 since January 2016. We know DC brushed motor is widely used for many different applications; from conveyor, AGV (Automated Guide Vehicle), mobile robot and combat robot. One of the reasons is the ease to use it, just connect the two terminals of motor to power res

Original URL: https://www.cytron.io/p-30amp-5v-30v-dc-motor-driver

*Thread Reply:* Jeff, I think this is what I was talking about the other day at the meeting about the issue of potenciometer, the motor and force, or are you talking about something different ..?

*Thread Reply:* I'll have to re-watch the video to see what exactly we were talking about.

*Thread Reply:* I watched the video. I'm not sure how this compares. You stated that you need to verify that the actual mechanical system is correct. This is exactly right. Various people tried to co fuse the issue. (Including me)

But yes, adjusting the pot on the driver board should control how much power (force) is applied to the motor. But controlling the PWM number form the Arduino sent to the motor would also control how much power (force) is applied to the motor.

Maybe we need to document this is more detail.

*Thread Reply:* Let's see if this is going to append this message or start a new message...

Yes, you need to determine correct functionality at a very basic mechanical level. Below ROS on Linux, and below a microcontroller. You need to verify the output voltage of a potentiometer by reading the voltage output. (assuming the potentiometer is wired as a voltage divider) This can be done with a multimeter (I.e., read the voltage) without any conversion using a microcontroller. If the votage is within an acceptable range, then you can move on to a microcontroller and verify the ADC reading that it is presenting. (I don't know how well this is translating to Spanish, Also, I have started drinking already...) But if it passes the voltage readings, then does it pass the the microcontroller ADC readings? Keep on mind that the range has to meet the range that is acceptable for the mechanical left to to right mechanical range. If any of this is does not match the accepted range then you have a problem. And yes, you have to control the applied power so you do not smash anything mechanically as you do your testing...

Hello Tawanda!

Hi Bob

Thanks for the welcome

I made a slight amount of progress. I am able to control my power steering motor position by turning an extra potentiometer. https://www.youtube.com/watch?v=MHd0v3Arq_g https://www.youtube.com/watch?v=EkU9EA50GnY

YouTube

} jws8675309 (https://www.youtube.com/user/jws8675309)

Power Steering Motor Test1

Original URL: https://www.youtube.com/watch?v=MHd0v3Arq_g

YouTube

} jws8675309 (https://www.youtube.com/user/jws8675309)

Power Steering Motor Test 2

Original URL: https://www.youtube.com/watch?v=EkU9EA50GnY

The weekly meeting is on Friday at 12:00 (Noon) Eastern time, 9:00AM Pacific time). About 12 hours and 45 minutes from whenever you receive this message. The Zoom meeting link is in the pinned message at the top of the General channel. (If I didn't screw this up...)

#general The ROS Agriculture meeting starts in 1 hour. The ZOOM meeting link is in the pinned message at the top of the General channel.

*Thread Reply:* @JeffS I'm on the run because it's my youngest son's birthday and I have a few kids on the farm. I think I'll be at the meeting; but I didn't want to go beyond asking .. this is a message that you made automatic ..? Because I'm interested in doing the same in other groups. On the other hand, I'm going to upload to the projects channel and generally what we were working on this week.

*Thread Reply:* This was not automatic. I just sent it at 1 hour before meeting. So people have a time reference. SInce we have several new people and the pinned message has not been updated with the schedule I thought I would post this today only.

*Thread Reply:* ahh, I was hoping that it could be automated because I have the same problem in the group in Spanish, I will continue to investigate it anyway. Thanks, see you in a bit.

*Thread Reply:* Recording from yesterday's meeting has been uploaded to the Google drive. The link is at the top of the General channel.

The diameter of the turning circle measured by the outer wheels of my lawn tractor while making a complete turn is ~13-14 feet. (~4-4.3 meters).  I know this because I executed cmdvel commands linear.x(speed) 0.8, angular.z(steer angle) -1.0 and +1.0 , distance (in meters) 80 and measured the results. Now, is that the "minturningradius" in teblocalplannerparamscarlike.yaml in /home/ubuntu/catkinws/src/lawntractor/lawntractor_navigation/config? Anyone? Ferris? @vinny ruia?

*Thread Reply:* link

*Thread Reply:* not able to make a turn less than the value stated... (not a fan of negatives) - so 4.3/2 or minturningradius = 2.15

*Thread Reply:* I think you are correct in that interpretation

*Thread Reply:* does this wiki page help? http://wiki.ros.org/teblocalplanner/Tutorials/Setup%20and%20test%20Optimization|http://wiki.ros.org/teblocalplanner/Tutorials/Setup%20and%20test%20Optimization

*Thread Reply:* I saw that earlier and wondered if it could be used in in the physical world.

*Thread Reply:* hmm I don’t see why not (just don’t launch stage)

*Thread Reply:* just be careful and keep your hand on the estop if it does some automatic tuning

*Thread Reply:* for future ref: https://wiki.ros.org/teb_local_planner/Tutorials/Obstacle%20Avoidance%20and%20Robot%20Footprint%20Model

*Thread Reply:* I was able to run a simple 2D Nav goal straight ahead. I adjusted teblocalplannerparamscarlike.yaml in /home/ubuntu/catkinws/src/lawntractor/lawntractornavigation/config in the tractor RPi: min_turning_radius: 2.2 # 14 feet diameter wheelbase: 1.27 # 50 inches front axle to rear axle xy_goal_tolerance: 1.0 # from 0.4 , just trying 1.0 yaw_goal_tolerance: 1.0 # from 0.2, just trying 1.0

*Thread Reply:* In other tests it was still doing a whole lot of reversing, so it would still be good to figure out how to reduce the tendency to go into reverse.

*Thread Reply:* not certain but you MIGHT be able to prevent the tractor going backwards by setting backwards vel to 0

https://github.com/ros-agriculture/lawntractor/blob/b9efa3345985e2135cf4db40a3e3265b8ad9d5a0/lawntractornavigation/config/teblocalplannerparams_carlike.yaml#L16

GitHub

ros-agriculture/lawn_tractor

Software for self driving lawn tractor. Contribute to ros-agriculture/lawn_tractor development by creating an account on GitHub.

Original URL: https://github.com/ros-agriculture/lawn_tractor/blob/b9efa3345985e2135cf4db40a3e3265b8ad9d5a0/lawn_tractor_navigation/config/teb_local_planner_params_carlike.yaml#L16

*Thread Reply:* cool, I'll give that a try

*Thread Reply:* Here is another thing to try. Section 6 on this page https://mowito-navstack.readthedocs.io/en/latest/step_5c.html says: weight_kinematics_forward_drive number Optimization weight for forcing the robot to choose only forward directions positive transl. velocities). A small weight (e.g. 1.0) still allows driving backwards. A value around 1000 almost prevents backward driving (but cannot be guaranteed).

*Thread Reply:* Found this which further suggests the weightkinematicsforward_drive setting helps backwards and forwards oscillation. Hoping to test tomorrow.

*Thread Reply:* for future ref: https://medium.com/@subodh.malgonde/building-an-actual-self-driving-car-53f67ca41566

Medium

Building an actual self driving car

How I built an indoor self driving car, 1/10th the size of an actual car

Reading time

4 min read

Original URL: https://medium.com/@subodh.malgonde/building-an-actual-self-driving-car-53f67ca41566

I have solved the squealing steering motor problem. The answer is to switch to the TimerOne library instead of using the AnalogWrite() function: https://www.pjrc.com/teensy/td_libs_TimerOne.html Or the TimerThree library would probably also work. The TimerOne says it works on Arduinos and Teensys. But the pin numbers that are available may vary.

It allows you to set the PWM frequency. I attempted to set it it to 20Khz (50us), I haven't verified it with a scope yet. But that frequency is not audible.

The other implication is that the pulse width range is 0-1023 instead of 0-255 on the AnalogWrite() command. I haven't tracked down any other side affects it may have.

I can give a demo between the two methods on the video meeting tomorrow.

Looking forward to your demo!

Hi, here it's the gist of my main.cpp code.. https://gist.github.com/rje1974/48f9bc3dc2685634ecaca59a7748b60b

@JeffS See if you can access this: https://gist.github.com/jones2126/bfebe7879e83ad2a7121bd7cace6f393

Yes, I can access that,

A tool to use for what we were talking about organizing the google drive .. This way you mount it on your ubuntu https://linuxhint.com/google_drive_installation_ubuntu/

Rowbot just sent me an email newsletter. You can see it here: https://mailchi.mp/050aaafc6357/field_report_cover_crop_seeding

*Thread Reply:* GPS antenna problems and navigation in high crops, I really hope that soon will be a common problem for all of us. Interesting application besides I hate cover crops

*Thread Reply:* This is something that @Javier has been working

*Thread Reply:* Yes Juan, in the case of corn the problem is complex, because when the robot is far fron the base station lost free path

*Thread Reply:* in the case of soybean it's most simple

*Thread Reply:*

*Thread Reply:* i put the gps at 1.5meters and works for soybean

*Thread Reply:* not for corn

*Thread Reply:* for sorghun is worst

*Thread Reply:* and if the day have high humidity is a problem too

I was just reviewing a couple of the original ROS Agriculture videos on YouTube. I see they are now set to "unlisted". So that is why they disappeared from the ROS Agriculture YouTube index page, but they still exist if you know the URL That doesn't answer if they will disappear completely some day.

You are going to have to be more specific...

Let me try this again... (Slack really pisses me off) Okay, I'm not quite sure what your confusion is. By searching my index that I posted: http://sampson-jeff.com/RosAgriculture/ grep ' Meeting ' ros-agriculture-youtube.txt > all_meeting This yields this. Not exactly, because I have more entries than what were posted. But I am going to ignore any further questions or statements from Bob...

I: 11302017 Community Meeting https://www.youtube.com/watch?v=mYeBaRp7DNc 12122017 Community Meeting https://www.youtube.com/watch?v=MSyXIZTZML4 01092018 Community Meeting https://www.youtube.com/watch?v=GrPFrgsnQsk I: 01162018 Community Meeting https://www.youtube.com/watch?v=BlDYHg9DOI4 ...

Oh, it won't let me post that . Because it is tons of information. So you have to do the search yourself.

I’ll be quiet. My apologies.

I downloaded few weeks ago all the videos of the ROS Agriculture youtube channel, i'm going to find a way to share the videos. Maybe upload again to youtube

*Thread Reply:* Maybe list them as private. Do we know which was the license that they originally had..? Because we really don't know who is the 'rigths owner' of the videos.. coming from an habitant of a country of the third world may be this sounds exaggerated but I don't want others to have problems..

*Thread Reply:* maybe in mega.nz

*Thread Reply:* have 50gb of free storage

I have a copy of all 250 meeting videos. I sent a copy to Al and to Bob. But it did not include all of the videos. You have a copy of everything.

I haven't decided if these videos should be reposted... Or where.

My videos used 90.1GB. Maybe it depends on which server the videos were downloaded from? Or the time of day?

I keep thinking we should start over. But r=there was lot of good data presented in the last 3 years.

I updated my index here: http://sampson-jeff.com/RosAgriculture/ I just added a new text file with the date 20210513 appended to it. As people have noted, if you view it in a browser you get extra characters. If you download the file and view it with a text editor then the extra characters go away. And view the readme.txt file for extra info...

Oh, I screwed up. I uploaded a bogus file. But I think I have it fixed now.

This was just posted to Home Brew Robotics email list: https://luci.com/wp-content/uploads/2021/05/LUCI-Safety-Report-2021-st.pdf Parts of this probably apply to any autonomous vehicle.

*Thread Reply:* I found the document very interesting and although I have to read more about the Luci standard, it seems to me that you are right is that it is something valuable and from where we can surely take as a basis for many instances, because being something that is directly related to a human, surely it takes much more into account to this than other products. Thank you very much Jeff ..

Someday I am going to have to figure out how to efficiently edit videos for upload. I have a couple of pictures and three videos from today. I may just upload the videos to YouTube and post the photos here.

I put my lawn tractor up on a jack stand to relieve the torque on the front end. Then I used the push-buttons on the Cytron 30-amp driver board to run the steering back and forth. It takes much more force than I expected when on the ground. And I decided that it requires almost exactly one turn of the steering wheel to go from left to right, These positions were not up against any hard stop, but close to full travel.

Here are photos of the setup: I need to figure out how to edit the size of photos in Linux...

I got my videos uploaded. Here is a hot from the front on low power. A shot from the back on low power. And a shot from the back on higher power. I need to get the frame attached solidly so it can withstand the torque (as opposed to me holding it). https://www.youtube.com/watch?v=hokrLYEN9XQ https://www.youtube.com/watch?v=9OzLPtF8f2k https://www.youtube.com/watch?v=uIF9KCYWabU

YouTube

} jws8675309 (https://www.youtube.com/user/jws8675309)

20210513 Power steering motor on lawn tractor #1

Original URL: https://www.youtube.com/watch?v=hokrLYEN9XQ

YouTube

} jws8675309 (https://www.youtube.com/user/jws8675309)

20210513 Power steering motor on lawn tractor #2

Original URL: https://www.youtube.com/watch?v=9OzLPtF8f2k

YouTube

} jws8675309 (https://www.youtube.com/user/jws8675309)

20210513 Power steering motor on lawn tractor #3

Original URL: https://www.youtube.com/watch?v=uIF9KCYWabU

Nice!

Videos from last two weeks are in the Google folder. They should be viewable in a few minutes.

Thanks!

2D Nav goal testing, tuning has not really helped. Still debugging. Will post video.

Hi Al. If you use the wheel odom source only, you have the same result for a rect line in small distance?

*Thread Reply:* @Javier Although I have wheel odom data I only use it to measure speed. Odom position comes from GPS via a conversion program. I will post the link.

*Thread Reply:* Juan and I have a sheet that lists our software components.

*Thread Reply:* I get odom by using this conversion program "gps_odom.py" https://gist.github.com/jones2126/0f98af1cad3ca73df3b0e759cc30dc3f

Did you fix the high publish rate that the URDF devices were publishing? I think they were publishing 10000/second.

*Thread Reply:* @JeffS Give me a hint, where am I looking for publishing rate? No I have only updated dimensions and tolerance related items.

*Thread Reply:* It was on the transform tree you would show. The devices listed in the URDF said they were publishing 10000/second. I would assume that would be controlled when you launch robotstatepublisher, or it is an entry in the URDF file.

*Thread Reply:* I am running my bagfile and captured this

*Thread Reply:* My urdf: https://gist.github.com/jones2126/ea6e3a2393ffc6f55a32ef9e22c2fe06

*Thread Reply:* Here is the launch file where two of the tf's are set up: https://gist.github.com/jones2126/e6334ce6d4fa1dbc79e83940216bc6b6

*Thread Reply:* for my future ref: http://www.zdome.net/wiki/index.php/Setting_up_the_ROS_Transforms

*Thread Reply:* Just a quick observation. Maybe you didn't notice the qx,qy,qz,qw fields. That is specifying a quaternion. I "think" if you supply 3 values it treats it as x,y,z. (or whatever the order of values to get pitch, roll, yaw). Then you won't have to figure a quaternion.

*Thread Reply:* I see in your launch file that the rate for robotstatepublisher is 30 (30 hz). But your transform tree says a rate of 10000hz. I'm guessing you simply have something broken. Possibly your "rate" value is being ignored. (Probably because your names do not match) Or something is just specified incorrectly.

I still haven't figured out why you don't just throw out your URDF and robotstatepublisher and just put the two static transform publishers in your launch file. If you want to kep your URDF, you are going to have to throw out the extraneous crap in the URDF. And figure out why the values from the URDF are not being used.

*Thread Reply:* This is suggesting 10K is sort of a default. Are you sure it is a problem? https://answers.ros.org/question/357030/problem-with-static-transform-publisher/

*Thread Reply:* Regarding the qx, qy, etc. there seem to be two formats for node pkg="tf". There also appears to be a node pkg="tf2_ros" which is more modern.

*Thread Reply:* That's true about the two different formats for statictransformpublisher. I just wanted you to be aware in the future when pull up that note that you are not required to supply a quaternion.

*Thread Reply:* Counting elements in my statement.....seem to be 6 <node pkg="tf" type="static_transform_publisher" name="map_to_odom" args="0.0 0.0 0.0 0 0 0.0 map odom 100"/>

*Thread Reply:* Okay "Bob", do what you want. But you have raised a very valid point. About publish rate, TF vs. TF2, URDF vs. statictransformbroadcasters. I will publish something ion a main channel so Slack doesn't hide it from everyone..,.

2nd Attempt at running 2D Nav Goal via RVIS

YouTube

} Al Jones (https://www.youtube.com/channel/UC_mdukR45X1FXZXJMlahr_g)

ROS 2D nav goal test using gps and move base flex

Original URL: https://youtu.be/VbNg6pe__iY

I'm start testing teb+move_base with odom from the wheels to debug the software

The first problem that i resolve is the accelerations limits. Maybe this parameter cause an oscilation

*Thread Reply:* Thanks for looking at it. Are you looking at teblocalplannerparamscarlike.yaml?

*Thread Reply:* Yes exactly

*Thread Reply:* So this is my current yaml file: https://gist.github.com/jones2126/3155dd13cf19ed10f8357a163b2a9333

and urdf: https://gist.github.com/jones2126/ea6e3a2393ffc6f55a32ef9e22c2fe06

*Thread Reply:* Try reducing acclimtheta=1

*Thread Reply:* If doesn'work increase the value

*Thread Reply:* And the acclimitx=1.5

One test could be reduce the max acceleration

Angular first

1st Attempt at running 2D nav goal (gps, rvis, move base_flex)

YouTube

} Al Jones (https://www.youtube.com/channel/UC_mdukR45X1FXZXJMlahr_g)

001 ROS 2D nav goal using gps - test 1

Original URL: https://youtu.be/eKMJyLI5iLg

Oh! i see the videos. Good test

The movements are very fast, maybe if the feedback is not too fast cause the problem

So... reducing the acceleration could be a possible solution..

Test using odom from wheels+tebplanner+rplidar

Avoiding obstacles

Using Plotjuggler to plot pose x, y from odom which is fed by gps. The result appears stable. When sitting for 90 seconds the movement ranges about 10 mm, so I'm thinking the gps signal and the conversion to odom is OK.

[Just to clarify. Slack bit me again and this will be a continuation of what I started...] [Again, Slack is fighting me with everything I type. So I will edit this in a text file and post it...] [I will post this in a main thread as opposed to a thread, so it is not hidden.]

Al has raised a very valid point. I see his transform tree says the values defined in his URDF file are publishing at 10000/second. That looked wrong to me. But he posted a link stating that it is normal if you are running TF2. https://answers.ros.org/question/357030/problem-with-static-transform-publisher/ (Point of order - TF2 was adopted with the introduction of Hydro (8 YEARS AGO!!!!)

They posted a couple of links (which is left up to the user to track down) that talk about using TF2 as opposed to TF.) So they are saying that you should not use: <node pkg="tf" type="statictransformpublisher" name="maptoodom" args="0.0 0.0 0.0 0 0 0.0 map odom 100"/> (which is TF) But instead you should use: <node pkg="tf2ros" type="statictransformpublisher" name="link1broadcaster" args="1 0 0 0 0 0 1 link1parent link1" /> The first method uses TF, which hasn't been used used (supported) for 8 YEARS!!!.... The second method uses TF2, which is the modern standard. Which has been the prefered (supported) method for the last 8 YEARS!!! It uses the "statictransform_broadcaster" from the TF2 package as opposed to the orignal one from the TF package. It also does not use the publish rate anymore. [this also has the format for quaternion as opposed to picth.roll.yaw, but that is different issue that you are supposed to inherently know].

So bottom line, don't use the original method any more for commnand line (or launch file) specification. Use the "new" method. (which may result in all of your topics indicating they are being published at 10000/second).

The second point, if you are going to use a URDF file for a physical vehicle, know it and understand it. The lines:

<!-- define base footprint (virtual link) --> <link name="base_footprint" />

Are broken. For two reasons. The URDF file uses XML formatting. A "<link" tag rerquires a terminating "</link" to terminate the block. So that breaks it. Secondly these is no code associated with the "<link name="base_footprint" />" block. So that breaks it. I don't know what the implications of these are in Al's case are, but it is wrong.

I assume you have to have a "link" and a "joint" for each object in a URDF file. And maybe other thing are required, but I don't use a URDF for a physical vehicle so I don't know the requirements.

I see some of the blocks are commented out (which was not immediately obvious). So I don't know if that particular URDF file is valid. (There is probably some form of "check this URDF file" avaialble, but I don't know what it is called.

I will go back to what I should be doing, making sure my taxes go out tomorrow. I can do a Zoom meeting or other direct interaction. But this random text interchanges on Slack are killing me...

[Oh, Slack bit me again by appending these togher...]

I think <link name=“base_footprint”/> is perfectly valid

And base footprint is connected to base link by L72 : https://github.com/ros-agriculture/lawntractor/blob/b9efa3345985e2135cf4db40a3e3265b8ad9d5a0/lawntractorsim/urdf/lawntractor.urdf.xacro#L72

GitHub

ros-agriculture/lawn_tractor

Software for self driving lawn tractor. Contribute to ros-agriculture/lawn_tractor development by creating an account on GitHub.

Original URL: https://github.com/ros-agriculture/lawn_tractor/blob/b9efa3345985e2135cf4db40a3e3265b8ad9d5a0/lawn_tractor_sim/urdf/lawn_tractor.urdf.xacro#L72

I do agree that likely using the tf2 static transform publisher is a good idea

Oh... It is because it is: <link name="basefootprint" /> Instead of: <link name="basefootprint" > which would require: </link> But as I said, I don't use URDF. So pardon my ignorance.

Not exactly ROS robotics, yet. My son has a bad back and can not twist around and open the seed gate of a tow behind broadcaster while driving his lawntractor. So I used a programmable remote FOB. I tried to follow the instructions but had problems. Seems one has to wait 25 seconds for the programming function to time out before it saves the selections. Any way it works. Now to mount everything on the tow behind spreader. I’ll order another fob unit for a kill switch on my project. Reason being that it can connect to other devices byway of I2C. I might need more auto shut down commands from future sensors?

*Thread Reply:* I understand that it would not be recommended for security reasons. any automated process that you cannot stop is dangerous

'Driverless tractors are indeed dangerous. So I’ll limit my current tractor augmentation to just using switches while I’m in the drivers seat,for now. I found a good candidate to act as my “crash dummy” while developing the necessary electronics. Ugly now, but who knows?

https://apple.news/Aq3518LVzTSuykX1HH2KE-g

apple.news

This Brain-Controlled Robotic Arm Can Twist, Grasp—and Feel — WIRED

Nathan Copeland learned to move a robotic arm with his mind, but it was kind of slow. Then researchers gave him touch feedback.

Original URL: https://apple.news/Aq3518LVzTSuykX1HH2KE-g

Anyone looked at this? https://docs.github.com/en/codespaces/about-codespaces I have tractor specific configurations I would like to easily share. I also have code that run on the Teensy's on the tractor. At the moment I have been using gist to share that and although that works I'm wondering if "codespaces" would be better. Hence the question, has anyone looked at this?

There was a question during the meeting today. What about steering?

Not ROS control but hey, gotta start somewhere. A programmable FOB activates the actuator which is connected to the seed gate/fertilizer gate. Pressing A on the FOB opens the gate. Pressing B closes the gate. The sliding limit switch more or less sets the application rate. BTW this is son Nate’s tow behind broadcaster spreader. Battery is behind the aluminum plate.

Nice video about something i thiink a lot and is about geting more from car parts especialy of the steering power motor.. https://www.youtube.com/watch?v=R6vzCr-iPoQ

YouTube

} Kevin Roscom (https://www.youtube.com/user/kevinroscom)

Toyota EPS hacking attempt 2

Original URL: https://www.youtube.com/watch?v=R6vzCr-iPoQ

Thanks Juan for the heads up! Nice video and very interesting. I’ll have to review it a couple of times in slow motion.

If you are up, there is a lunar eclipse in 4 hours: https://www.timeanddate.com/live/eclipse-lunar-2021-may-26

timeanddate.com

LIVE Stream: Total Lunar Eclipse May 26, 2021

Watch a LIVE broadcast of the Super Full Moon turn red during the Total Lunar Eclipse on Wednesday, May 26, 2021.

Original URL: https://www.timeanddate.com/live/eclipse-lunar-2021-may-26

[Will this get appended to my previous message?] I attended the Home Brew Robotics ROS Zoom meeting tonight. They went through converting a ROS1 Python program to ROS2. It was interesting to see what changes were needed. Mainly it was replacing rospy with rclpy. And each call to rclpy needed to be converted to the new new format.

i use mcedit

https://midnight-commander.org/ the mc have the mcedit editor

vscode and micro

Thanks guys!

Pods detection on soybean. Model trained using yoloV5. https://github.com/ultralytics/yolov5

https://discourse.ros.org/t/first-beta-of-ros1-integration-for-farming-simulator19/20625

ROS Discourse

First beta of ROS1 integration for Farming Simulator19

Hi all, We’ve just made FS19_modROS available to the public. It’s all work-in-progress but if you are interested, feel free to take a look. Have fun!

Reading time

1 mins 🕑

Likes

3 ❤️

Original URL: https://discourse.ros.org/t/first-beta-of-ros1-integration-for-farming-simulator19/20625

My FOB kill switch works. Once it’s on the Jeep I’ll use bigger wires. The fob relay controls a heavy duty relay. The toy motor is there temporarily to test the concept. It will be replace by the Jeep’s drive motors.

http://wiki.ros.org/ros_arduino_bridge

This is the library that i'm using to control the motors of the wheels and the steering. Works very well and stable using an arduino UNO board connected using the USB to a computer with ROS. In my case like as said in the meeting i used an odroid board with 4GB of RAM

https://www.hardkernel.com/shop/odroid-n2-with-4gbyte-ram-2/

This is the tutorial I followed to get the Blue Pill to work like any other Arduino. I didn't test it alot, but I can directly upload code from the Arduino IDE over USB, and I could do a print statement and it would show up in the Arduino terminal screen. I didn't check to see how many Arduino libraries are available. https://www.youtube.com/watch?v=Myon8H111PQ

YouTube

} Caleb Marting (https://www.youtube.com/c/CalebMarting)

How to Install the STM32 USB Bootloader, New Version! (Blue Pill F103C8T6)

Original URL: https://www.youtube.com/watch?v=Myon8H111PQ

Here is a simplified diagram of what I was trying to explain during the meeting. This phase sole purpose is to determine the FOB control range.

thanks

@JeffS , did your Dan Foss tractor have its kill switch before or after the on/off key?

I'm not sure exactly what your question is, or how it relates to your Jeep. But we put our relay between the battery and the ignition switch, If the relay opens it was like turning the key off. So the engine would die and all of the auxilliary electronics on the tractor shut down.

*Thread Reply:* I don't recall any suggestions about moving eStop circuits. I just told you what we did on the eStop circuit on the Danfoss project.

*Thread Reply:* I should have been more accurate. I should have said, I’m going to borrow that Dan Foss idea.

*Thread Reply:* Thanks for the clarification. I’ve relabeled my circuit as “ interrupt circuit” the FOB toggles on....or toggles off ...the rear axle motors, at this point.

Camp Peavy is starting a robot similar to Bob's new jeep. Here is a link to his description. https://groups.google.com/g/hbrobotics/c/jiGSzRuWG-U/m/9TOaWjmBAwAJ

You guys are likely better linux admins than me so maybe you know this, but thought I would pass along I was trying to update something on my RPi ($ sudo apt-get update); I had the error, "...signatures were invalid: KEYEXPIRED 1622248854..."; After running $ apt-key list I found an Open Robotics key had expired causing an issue with http://packages.ros.org/ros/ubuntu xenilease

pub   4096R/AB17C654 2019-05-30 [expired: _**2021-05-29**_] uid                  Open Robotics &lt;<a href="mailto:info@osrfoundation.org">info@osrfoundation.org</a>&gt;

After research, the fix was $ sudo apt-key adv --keyserver <a href="http://keys.gnupg.net">keys.gnupg.net</a> --recv-keys AB17C654

I'm assuming your key will be different than AB17C654

Try running apt-get install -f

sudo apt-get install -f

*Thread Reply:* Now looking in to how to solve this....

*Thread Reply:* sudo dpkg-reconfigure -a

*Thread Reply:* Did you mean --force?

*Thread Reply:* Sorry yes with then force argument

*Thread Reply:* Its a very powerfull command i solved a lot of problems with dpkg

*Thread Reply:* Then run again sudo apt-get install -f

*Thread Reply:* sudo apt-get update

*Thread Reply:* Maybe this applies: https://groups.google.com/g/hbrobotics/c/uWisDR7quYE/m/pjxrbyB1AwAJ

I got the fob interrupt working. It has a range of 300 + feet BUT the slightest blade of grass will interfere. So my lesson is to be careful, it may not work when I need it. I have to have the FOB receiver high AND totally unobstructed, if possible. I am satisfied at this point. All I wanted at this point is some control within a few feet.

https://youtu.be/j7CzFeHqf0s

YouTube

} FS Farmer (https://www.youtube.com/channel/UCBd6VvmjakD4yfAgpOoYHNA)

Amazing Modern Agriculture Machines Tractors in Action

Original URL: https://youtu.be/j7CzFeHqf0s

Simulating crops in gazebo

*Thread Reply:* This is a development for the project you are working on ..?

Hi Juan, yes. We working to build another robot, the simulation helps in the design and we testing all using gazebo before build

*Thread Reply:* NICE, I really think it is a very good business model.

Was last FRIDAY’s meeting recorded?

*Thread Reply:* Yes, I have to convert it and upload it still.

It has been really hot the last few days. So I decided to move downstairs and play with some YouTube publishing software. Here is a test result from today: https://youtu.be/4fW7nESS6BQ I am fighting with the audio. Is it usable? Too loud? Too soft?

YouTube

} jws8675309 (https://www.youtube.com/user/jws8675309)

OSB software testing

Original URL: https://youtu.be/4fW7nESS6BQ

Looks and sounds good here!

I have come to the conclusion that the green screen in Zoom is FAR superior to OBS software. In Zoom, you just turn it on and it works. In OBS there are many parameters you can change, and none of them work very well. Maybe I will hang a neutral back drop and not worry about it.

This description at the top of this page may fix your expired keys, if people are still having problems... https://discourse.ros.org/ This was just mentioned in the Home Brew Robotics ROS meeting...

discourse.ros.org

ROS Discourse

The ROS discussion forum

Original URL: https://discourse.ros.org/

*Thread Reply:* Hi Jeff, broken links..

*Thread Reply:* Which links are broken? I can get to https://discourse.ros.org/ I can execute curl -s <https://raw.githubusercontent.com/ros/rosdistro/master/ros.asc> | sudo apt-key add -

*Thread Reply:* ahhhh ok

*Thread Reply:* FYI: I didn't actually read it. Somebody said it fixed their problem. When I just skimmed through it now, it sounds like they may have it fixed anyway.

Finished the 4 teensy tutorials using 3.2 board. In tutorial 4 https://www.pjrc.com/teensy/tutorial4.html they used a simple 10k pot.

So now I want to experiment with maybe 2 joy sticks like these https://www.adafruit.com/product/3102?gclid=Cj0KCQjwzYGGBhCTARIsAHdMTQyrLOql8cxIEMF175jRD8JHfWouM4K57wC9q97ZjztEN5-bEvc124aAoZkEALwwcBf

JoyStick #1 would control forward/reverse ...up/down on the stick And control steering left/right on the stick

Anyone using these? Any recommendations for a different product?

I’ll go from analog to pwm into my cytron motor controller. I’ll have to figure out how change the directional on the motor controller.

adafruit.com

Mini Analog Joystick - 10K Potentiometers

Sometimes a simple analog control device can be the perfect tactile solution for your project, but they can be surprisingly hard to come by. Luckily we&amp;#39;ve found a low cost, quality Mini ...

PRICE

$19.95 USD

STOCK

33 IN STOCK

Original URL: https://www.adafruit.com/product/3102?gclid=Cj0KCQjwzYGGBhCTARIsAHdMTQyrLOql8cxIEMF175jRD8JHfWouM4K5_7wC9q97ZjztEN5-bEvc124aAoZkEALw_wcBf

Those joysticks are almost identical to the ones I used.

Al points out that you may want to disable the spring-return-to-center (for one or both axis). Some joysticks have a small lever that enables or disables the spring. Or, on the lager ones, you can probably use small needle nose pliers to pop the spring off and that disables it. (Or maybe not...) But don't create a situation where you may leave a dangerous situation if you take your hand off the joystick. I.E., you probably want forward speed to always return to zero if you take your hand off. The steering axis can be your preference.

If you are just experimenting, you can get cheaper/smaller joysticks. https://www.adafruit.com/product/512 https://www.adafruit.com/product/245 https://www.amazon.com/s?k=thumb+joystick+arduino&ref=nb_sb_noss_1 https://www.mpja.com/Mini-X-Y-Joystick-10K-with-Mushroom-Knob/productinfo/34792+MI/ https://www.ebay.com/sch/i.html?_from=R40&_trksid=p2380057.m570.l1313&_nkw=joystick+arduino&_sacat=0 If you get a small one, you want one soldered to board. It will be be easier to mount and easier to connect.

Another option is a 3 axis joystick: https://www.mpja.com/X-Y-Z-Axis-Motion-Control-Joystick-10K-Resistance/productinfo/34791+MI/ But to me those look like they would be hard to use.

And of course the other approach is to use a USB joystick through your ROS computer. You can get hardwired joystick/gamepad or get a wireless gamepad to connect to your ROS computer. Al switched over to a wireless gamepad directly into his ROS computer on his lawntractor.

That has the advantage that you are running through the same path that the autonomous path will be using. The disadvantage is that you have to have the ROS computer booted up and running ROS to get this to work.

The main advantage to connecting a joystick directly to your Teensy is that you can drive you vehicle directly with ROS not running or even without a ROS computer connected. My remote control bypasses ROS completely so I can turn on the main power switch on the robot and then turn on the remote and start driving. I find that very handy. But I have less incentive to get my ROS based control working... 🙂

In the long run, it would be handy to have both types of joystick available. So that is something to think about in the future.

For now I want to bypass ROS. I decided against the Bluetooth controllers because eventually I want to use RF with the joysticks. I like the alternatives you suggested for the small size. I just don’t know how long they would last with repeated use out side.

Another thing to consider with any joystick control...

Have a way to qualify the control. Have a physical switch to switch between manual control or automatic control. (Which doesn't solve the real problem.)

The real problem is that you may accidentally bump a joystick and your machine will move unexpectedly.

So you can add a "dead man" function. Typically an extra puch button on your remote control device. You must hold down the pushbutton for the joysticks to be honored. If you are not purposely pressing the dead man button the the machine can't move unexpectedly.

Or another approach is to put a gaurd around your joysticks. (or recess them down into your box so you have to reach inside with your thumbs to push the joysticks. Which is what I did because I was copying someone else's product.)

Or combine two or three of these methods...

Here Kyler demontrates what happens if you bump a joystck and you did not implement a dead man button: https://youtu.be/TatrSXJiq0I?t=390 You can watch that video from the beginning to see if first attempt at a remote control.

YouTube

} Kyler Laird (https://www.youtube.com/c/KylerLaird)

portable joystick system

Original URL: https://youtu.be/TatrSXJiq0I?t=390

https://link.medium.com/paJfJnSwXgb

Medium

Connect PS3 Joystick to Raspberry 4b with ROS

Example of usage of the packages teleop_twist_joy and joystick driver for ubuntu mate to connect a PS3 joystick and interact with real or…

Reading time

7 min read

Original URL: https://link.medium.com/paJfJnSwXgb

If you want to use ROS with ps3 joystick I wrote an article about how connect to ROS with a raspberry pi 4. I have an image of the sd card with ubuntu and the ROS packages installed and ready to use to share.

Thanks Jeff and Javier for the suggestions ! I will remember them.

https://agrirobotics.wordpress.com/

Agricultural Robotics

} fkuelbel (https://agrirobotics.wordpress.com/author/fkuelbel/)

Call for Applications

The deadline for applications is June 15, 2021. The Topics Agricultural production of crops as a source of food and energy faces enormous challenges: on the one hand, yields must be increased to meet the needs of a growing global population, on the other hand, the available arable land is shrinking, the effects of climate change are already noticeable in many places, and the ecological footprint of crop production must be reduced. Digital technologies such as agrorobotics in particular have the potential to make a significant contribution to overcoming these challenges through optimized observation, analysis and intervention. In the context of the Summer School, these topics will be addressed and worked on. Workshops: Introduction to Machine Learning &ndash; Prof. Dr. Ribana Roscher (University of Bonn)Deep Learning for Applied Image Classification &ndash; Prof. Dr. Chris McCool (University of Bonn)Basics of Graph-based Simultaneous Localization and Mapping (SLAM) &ndash; Prof. Dr. Cyrill Stachniss (University of Bonn)Crop Phenotyping for Increased Sustainability &ndash; Prof. Dr. Achim Walter (ETH Z&uuml;rich)A Practical Tutorial in Robot Perception and Control &ndash; Prof. Dr. Marc Hanheide &amp; Prof. Dr. Grzegorz Cielniak (University of Lincoln)Analyzing Temporal Visual Data and Forecasting &ndash; Prof. Dr. J&uuml;rgen Gall (University of Bonn)Economics of Robotics and Sensing Applications in Agriculture &ndash; Prof. Dr. Jan B&ouml;rner (University of Bonn)Target Group Advanced Master&rsquo;s students (second year)Early doctoral students (first year)The Event The International Summer School will take place online, consisting of workshops, pre-recorded video talks, discussions and networking events. The event will be realized via an online conferencing tool. The summer school is free of charge for all participants and presenters and will be taught in English. All nationalities are welcome! How to Apply Submit your application form via e-mail to &nbsp;&nbsp; This includes: Your name, current academic status, name of your supervisor, field of researchA copy of your Master&rsquo;s certificate or current transcriptAn abstract of 300 to 500 words outlining your motivation to participate in the summer school and how its topics connect to your research interestsDownload application form ECTS You may earn 3 ECTS for continuously participating in the summer school as part of the PhenoRob Graduate Training Program. Please make sure to check whether it is possible for you to earn credits as part of a different PhD-program. Contact Please contact the PhenoRob Cluster Office via should you have any questions or concerns. We are looking forward to your applications! Download CfA

Original URL: https://agrirobotics.wordpress.com/

I hate to impose my bias here. But if you were just bitten by Microsoft's latest update non-sense (Microsoft News Feed), here is a way to fix it... https://www.youtube.com/watch?v=_CpCkFa5Gi0

YouTube

} hirudov (https://www.youtube.com/c/Hirudov)

How to Remove the News and Interests Widget on the Windows 10 Taskbar with Local Group Policy Editor

Original URL: https://www.youtube.com/watch?v=_CpCkFa5Gi0

My power is fluctuating. So I may not make the meeting tomorrow (later today). Either because my alarm clock may not go off or simply becuase I may not have power. We'll see...

Hello, I stopped by the meeting and I see that it was complicated for everyone .. I leave you my update of the week ..

1. After a lot of turning I realized that the USB of my rpi burned or something like that. I didn't feed my teensy and it drove me crazy. So new rpi. I understand that I can possibly rescue her by pulling a fuse.
2. We end up with something similar to the system control script from arduino. It is still missing but let's say this could be a very bad first version, I went back to work with my "apprentice" after like four weeks.
3. I've been thinking a lot about Jeff's comments on how the ROS systems, Al's and mine are structured. He is right in several respects but mainly that it is a mistake to go around calling things from base stations. I'm going to see how much I can go from my base to the robot and simplify ..
4. I've been looking at my main.cpp and seeing how I can comment on it better

Al said he is going to be late. So Bob and I are waiting for the meeting to start.

I'm done waiting. I'll see people next week.

I got stuff to do outside before the humidity gets bad. C u next week.

So trying to figure out my steering more than just sending -1 to 1 in cmd_vel.

*Thread Reply:* I have two sensors to detect the steering angle. One is analog and one is a magnetic sensor. Readings for full left, straight and full right are

*Thread Reply:* So if I use a protractor on the angles it looks like

*Thread Reply:* @vinny ruia or @Javier or anyone that wants to comment about movebaseflex.... When movebaseflex sends cmdvel twist message, what basic instruction is it asking the low level steering controller to execute? At the moment, a cmdvel angular z of -1 would equate to a right turn at 156 degrees and cmd_vel angular z = +1 would equate to a  left turn at 162 degrees on my tractor

*Thread Reply:* I guess I should add, it appears the lawntractor package is expecting movebase/cmdvel to send values between -0.3 and 0.3 and then remaps those values to -1 and 1, but the basic question is still the same, when movebase/cmdvel sends values between -0.3 and 0.3, what physical behavior is it expecting the low level steering controller to deliver? (e.g. If movebase/cmd_vel sends -0.3 is that an expectation the physical steering controller turn right 0.3 radians/second?)

GitHub

ros-agriculture/tractor_teleop

A simple ROS teleop drive system for car-like robots. - ros-agriculture/tractor_teleop

Original URL: https://github.com/ros-agriculture/tractor_teleop/blob/master/src/tractor_teleop/cmd_vel_mux.py

*Thread Reply:* It seems a little fishy to me that you have 156+162=318 degrees of steering. From you protractor photos, may you have 30 degrees one direction and 25 degrees the other direction. But I don't know what that is you are measuring there.

By definition, ROS planners output rotational velocity for steering.

But... if you are using the TEB local planner, and you set the parameter to use steering angle instead of rotation velocity, then it will be an angle.

*Thread Reply:* Technically I've used the wrong language. Ai would be 162-90 or 72.

*Thread Reply:* You may want to be skeptical about the existing code. To me it looks like a hack. Probably to get the Stage simulator to work.

The code in the in the command velocity mux you are using is articially limiting the output of the TEB planner to +/-0.3. SInce you told the TEB planner to out an angle instead of rotational velocity then the value is interpretted as an angle of +/-0.3 radians. One radian * 0.3 (573. degrees * 0.3 = 17.19 degrees) So right off you are artifically limiting the output of the TEB planner to +/-17 degrees. That seems like an awfully small angle to me. It should work if you are only attempting to drive straight lines.

Then the code blatantly multiplies that by 3 (the 0.3 to 1.0 translation). So the value that was clipped at +/-17 degrees is scaled to your full value. Seems to me that will cause massive oversteer.

But the bottom line is. You have to know if you want your controller wants to see an angle or it wants to see rotational velocity. (Hint: you told it to expect an dangle because the Teb planner was told to output angles.

So, by definition, you want your low level controller to expect and input as an angle (in radians) and scale that to your actual steering angle.

Currently you code is scaling an input of +/-1.0 to -14003 to 13683. (Or maybe you are doing +/-1.0 to 1013 to 17.) Instead scale the actual measured angle to your sensor value. Let's assume you have a more realistic value like +/-30 degrees (0.5236 radians). So scale +/-0.5236 to -14000 to 13683. [Regardless of what your your panner is putting out or what your joystick is putting out. Because by definition, the input value is a an agnle that should match your physical steering.]

If you get that working, then the TEB planner is happy, it just outputs angles and expects your vehicle to listen and do it. You may have to recscale the joystick, either hack the joystick node, which was probably hacked already to get +/-1.0 (which is probably based on teleop_joystick) [It may have input parameters to do scaling for you.]

Or hack the command velocity mux to scale it. Or create a node that goes between the joystick node and the command velocity mux node.

Now I will stop and look at what you just posted...

*Thread Reply:* Al said "Ai would be 162-90 or 72." Can you define that a little harder. Do you mean steering of 0 is 0? And fulll left is -90? And full right is 72?

*Thread Reply:* My protractor measurement of 162 for a full left turn is an obtuse angle best represented in this diagram as Ai +90 degrees

*Thread Reply:* I never tried to calculate an Ackerman steering angle.

*Thread Reply:* Another issue is that Ackermann is not linear. Sometimes in descriptions people will talk about "an imaginary center wheel". That center wheel is the actual angle you request. But both the left wheel and right wheel will be different angles. The inside wheel turns sharper and the outside wheel will be less that the center wheel. I am sure there is a transfer function for that, but I don't what it is. Maybe I will go look for it.

*Thread Reply:* I think somebody is still confused on your measured angles (maybe me). I am assuming you have a bars on the ground. One reresents the centered line and the other is parallel to your tire. (I.e., the two lines defining Ai above.) Looking at your protractor for the left side the protractor says either 30 degrees or 152 degrees. (And 180-152=28) What does 90 have to do with it?

*Thread Reply:* @vinny ruia I have two questions.

1. Should I add to my low level steering controller reading the current speed of the tractor in order to be able to calculate the angle to achieve the correct radians per second expectation from move_base? I think Jeff is pointing out we need to steer to a radians/second target. At the moment, speed is not an input to my lowest level steering controller.
2. There is what I will call a clamp in cmdvelmux.py to limit cmdangle.z +/- 0.3. Do you know why? My tractor has a turning radius of ~2.15m. At 1.5 m/s I think that gets me to 0.7 rad/s. I'm going to assume since I set minturningradius in the yaml file that movebase will call for steering up to that point.

*Thread Reply:* HI Al, in my case I calculated in the controller the turning radio of the wheels. If the move_base package sent an angular velocity of 0.3 means 0.3 radians per second and depend of the geometry of your robot you need to calculated the ICC to respond to the requirements of the move base.

*Thread Reply:*

What is the quickest way to read data from a USB/serial port and publish as a ROS topic?

I want to plug in a Sparkfun Razor IMU (probably with the Sparkfun code). Set it to put out (at least) the raw (degrees/sec) values from the gyroscopes. Then I want something quick and dirty to read the strings of data and publish them as a ROS topic so I can record them in a bagfile.

Yes, I could setup a full ROS driver for an IMU. It does publish rotational rates in the IMU message. But I don't know if those are raw unadulterated values, or if somebody played with those (in their infinite wisdom). And my experience is that it would take days to get REAL values from any IMU.

*Thread Reply:* Jeff, good morning.. Nice sensor.. Did you see this video..? https://www.youtube.com/watch?v=we5FBjllVn0 and this driver.. https://github.com/mitll-ros-pkg/razor_imu_m0_driver

YouTube

} JetsonHacks (https://www.youtube.com/c/JetsonHacks)

Jetson RACECAR Part 9 - Razor IMU ROS Software Install

Original URL: https://www.youtube.com/watch?v=we5FBjllVn0

GitHub

mitll-ros-pkg/razor_imu_m0_driver

Razor IMU M0 ROS Driver. Contribute to mitll-ros-pkg/razor_imu_m0_driver development by creating an account on GitHub.

Original URL: https://github.com/mitll-ros-pkg/razor_imu_m0_driver

*Thread Reply:* Ok

I didn't find a quick/easy way to publish a serial stream. So I hacked one together. It creates a topic like this:

```data: "11126903, -0.85, 0.37, 0.00"

data: "11127003, 0.61, -0.12, 0.43"

data: "11127103, -3.17, -0.37, -0.55"

data: "11127203, -0.61, 0.37, 1.40"

data: "11127304, 0.79, 1.10, 4.57"

data: "11127403, -8.72, 0.37, 20.06"

data: "11127503, -2.99, -4.76, 32.07"

data: "11127603, 14.45, -2.62, 39.57"```

I assume most people aren't subscribed to Home Brew Robotics Group and probably won't be. So here is a quick update on Camp Peavy's plastic car robot: https://groups.google.com/g/hbrobotics/c/jiGSzRuWG-U/m/9TOaWjmBAwAJ https://photos.app.goo.gl/Qh9i4nj63wHx3d1R8 https://youtu.be/NhnkrQyU0Us https://photos.app.goo.gl/rze1biRVWu3W39vQA https://youtu.be/AdPY_YmWC3k https://photos.app.goo.gl/FNE3WCXTYx44phgr8 https://youtu.be/xhwEK7FCC-8 This leaves out a lot of the commetary, but if you want more info you can go track it down. I probably won't post any more updates on this.

The weather is supposed to cool off in the future. So I was planning to configure my robot today so I can take it outside and do some steering angle verification. But my Internet provider decided to make my life a living hell the last couple days, so I haven't gotten anything done. But I need to configure the wheel odometry, the dual GPS and the gyroscope rotational velocity so I can verify if I know what I am doing with steering angle...

But instead I have been fighting with the decision of DSL, cable, wireless, cell modem in the brief times I have Internet connection... I was was wondering if I can run my whole house Internet on a cell phone modem... So I haven't gotten anything done.

Can’t make it today. Got grandparent duty this weekend.

I tried to get uBlox ucenter to run under Wine on Linux. It wasn't obvious how to translate the COM port names to make it work.

I reverted back to ucenter under an actual Windows laptop. I was able to load both configuration files. (one for the main RTK receiver and the one for the "moving base line" "rover" receiver on the robot. So the main receiver generates the precision Lat/Lon position and the second receiver gives me a static heading (if I am moving or not). I need to get it put back on the robot to verify it works there.

I need to take it outside and verify I can get GPS data, IMU data and wheel odometry data and log the data. Then I need to verify I can get "recovered" "rotational velocity" from IMU, wheel odometry and from GPS. If I can get all three sensor to agree, then I have a way to verify the actual path/rotation of the vehicle. Then I can attempt to give it a steering command (either steering angle or rotational velocity) and verify I can get the input to match the recovered output.

Maybe I can try that tomorrow...

But due to my laziness, I may not do anything tomorrow...

So I got things configured and I went outside to make a sample bag file. I wanted to know where it puts trhe bag files, is the data useable and what other problems I will have to fix.

I found out a cable was in the way and I couldn't easily wedge my laptop into place. With my GPS receivers inside of a box I have no idea what the status is. But I drove around anyway. So now I have a bag file.

For some reason rqt_plot will not work on my kinetic laptop. Plotjuggler will not load on my kinetic laptop.

I copied my bagfile over to the noetic laptop. I loaded Plotjuggler. I can't figure out how to do an X/Y plot. It just doesn't seem to work.

My topics that are just a string of numbers are not recognized. So I have to decide how to fix that.

So now I have several things to look at and try to fix...

So out of desperation,,, I shutdown my Ubuntu 20.04/Noetic laptop and restarted it. Now I can do X/Y plots...

After starting Plot Juggler multiple times, I have gotten 3 different initial screens. Now it has settled down now. I am able to plot the topics that have individual fields. Here is in an X/Y plot of odometry and GPS. The odometry is rotated since I didn't think to align my vehicle before I started ROS. The only thing I could get rotation velocity was odomentry. Looks like I am getting 0.8 radians/second when driving in the tight circle. I need to work on the other two sensors to present rotational velocity. (Or I can hack on the strings to separate them...) 3 files

I looked at my numbers some more. Rotational velocity from RTK GPS (I took the change of heading per sample, scaled to degrees/sec, divided by the update rate and conveted to radion/sec.) It says approximately 0.35 rad/sec (full left or full right) I plotted rotational velocity from my wheel odometry message and it is approximately 0.8 rad/sec. One, or both, are wrong. I then plotted the heading from GPS and Odometry (the blue curves). They should look the same but they don't. The one from GPS is probably correct. The one from odometry is obviously wrong. (But I looked at an old plot of heading from odometry and it looks right.) Maybe it is in my interface node where I take values from the microcontroller and build/publish and odometry topic.

I didn't plot the rotational velocity from my IMU. It appeared to die shortly after I started the bag file. I think the cable was pulled loose.

Looking at the path from odometry I can measure the circle created as I drive in a circle. This is the path that base_link is moving. It is approximately 2.0 meters diameter driving in either direction. I can measure the GPS, but it is in Lat/Lon and I haven't converted it to meters.

I think I solved the mystery of the odometry heading value not looking like the GPS heading value (in the plot above). It looks like my ROS node which takes raw data from the microcontroller and builds the odometry message is wrong. I am simply pulling input fields and plugging them into output fileds. I probably picked a value for the quaternion and plugged into an inappropriate field in the new odometery message. So I can probably fix that.

I still haven't figured out why odometry say a rotational velocity of 0.8 and GPS says 0.35. Maybe if I run again with the IMU I can see it matched either one. If so I will know which one is wrong. It it doesn't match either of them then I am totally baffled for the moment. But it raining now so I may try the IMU tomorrow.

*Thread Reply:* Good work Jeff!!

*Thread Reply:* Perhaps the difference between the gps and odometry from wheels are related to the calculation to translate the ticks from the encoders to odom.

*Thread Reply:* Hey, that is a good point. I can plot odometry linear speed against the GPS linear speed. I'll go try that.

Oh, I forgot to send that about two hours ago. I ran my IMU it looks fairly clean it thinks my rotation velocity is about 60 degrees/second. The GPS currently thinks that was about 20 degrees/second. I'm tired of playing with it today...

So here is the speed check between GPS and wheel odometry. The ground speed from navpvt is mm/s. The linear x in odometry is m/s. So they both look approximately the same speed.

Maybe things aren't off as far as I thought. The odometry is indicating about 1 radian/sec. The IMU indicates about 60 degrees/sec.

60 degrees/sec is approximately 1 radian/sec. So maybe they are close.

But... the GPS is still telling me 20 degrees/sec. So I have to figure out the GPS scaling.

I modified my gyro ROS logging to output a single Float32 instead of 3 values in a string. So I can now plot that directly in Plot Juggler. I also scaled it by converting from degrees/sec to radians/sec and multiplied by -1 to flip it. My first run today was baffling. I drove in a circle one direction and then drove the other direction. And the second circle was at a much faster forward speed. Only the first circle looked right. When I plotted the rortational velocity and forward speed, I see the bad circle had a forward speed of around 2 m/s. That is way too fast. And apparently the tires were slipping on my dry grass.

Then I drove around again at a more controlled speed. And this time I drove in several size different diameter circles. These plots look much better. The odometry path looks close to the GPS path. But one is rotated and maybe flipped. The GPS has at least one jump... The rotational velocity from the odometry matches the rotational velocity from the gyroscope. It tracks well enough to give me confidence that I did it right. And it also tells me that I should not go above 1 m/s forward speed...

Now I have to figure out how to convert the GPS lat/lon to offset in meters in real time. Then I can plot those two on top of each other in Plot Juggler.

At one time we discussed how you know if your tires are slipping. The fact that my two rotational velocities don't match above indicates that something something is wrong. You could have a separate node to watch for these things. Then you have to think what to do about it once you detect it.

Amazing work Jeff!!

Nice job!

Really great

Wohhhh

I'm getting closer at converting my GPS heading to rotational velocity. My biggest problem was I was dividing by 10000 instead of 100000. I take the heading output and find the delta from current sample to last sample: Delta=Heading(T1)-Heading(T0) Then divide by large number to scale to actual degrees: Degrees=Delta/100000 Then convert degrees to radians by dividing by 180/PI: Radians=Degrees/(180/PI) Then account for time units by dividing by 5 samples/second. But that is off by a factor of 2. So I divided by 10 instead. Then threw in a * -1 to flip it to match the gyroscope put. Rotational Velocity=Radians/10*-1

So I still have a "off by 2 times error". I don't know why. Maybe the GPS is not sampling at 5 Hz, like I thought I was. Maybe the GPS has multiple sample rates internally? So I have to track that down.

*Thread Reply:* Hi Jeff, can you record a bag file with all the topics? The command is rosbag record -a. Then you can use rqt_bag to upload the recorded file and see the topics at the same time. It's a good way to see if there are a problem with the sampling. If you share the bag i can check it

*Thread Reply:* http://wiki.ros.org/rqt_bag

*Thread Reply:* Great work Jeff!

*Thread Reply:* When you obtain the heading from the GPS, the Delta is represented by two measurements (T0, T1), perhaps the problem come from there . I'm not sure...

*Thread Reply:* I am recording all topics to a bagfile. I can plot the topics directly from the bagfile with Plot Juggler (the screen shot above on the left). I can also use either of two different programs to break the bagfile up into individual CSV files. Then I can load CSV files into Open Office Calc (or Excel) and do calculations and plots. (the screen shot above on the right)

I can't figure out how to get Plot Juggler to plot bagfile values AND additional CSV data at the same time. (That may cause an issue of sample rate of the CSV vs. bagfile topics.)

Apparently I can do extra calculations in Plot Juggler. But I haven't tried to understand that yet. It looks like it uses LUA.

Another problem is that I have to use two computers to make this work. I record the original bag file on the robot computer. That computer will run the bagfiletocsv programs. But I can only get Plot Juggler to run on the other computer. So I am always copying files back and forth. It gets tedious...

After doing about 6 steps to convert heading to rotational velocity, it could be any step. I'm just missing a scale factor of 2:1 somewhere. If I just ignore the fact that I don't understand, it is giving me the correct data. Eventually it may become obvious.

I’m having trouble centering my joystick to zero. I’ve fiddled with rotating the pot And my numbers in software. Any ideas?

I’m just focusing on the forward/reverse. Btw the pots are 10k.

Could you be a little more specific about what expect out of your joystick or what you are expecting it do.

After a couple of days of fighting with Python I have a ROS node that subscribes to the uBlox F9P /gps1/navrelposned message and converts the static heading into a rotational velocity as it runs. I plotted the gpsrotvel with gyrorotvel and they appear to track each other. I would have plotted the wheel odometry rotational velocity, but my robot failed to log any data for it when I drove it around. I have to do some playing around to eliminate those spikes. They are caused by the rollover between 0 and 360 degrees.

*Thread Reply:* Excellent result Jeff. So your rotational velocity from the gps heading match with the measurements of your gyrocope.

*Thread Reply:* did you test the navigation stack?, for example: if you graph the angular and linear velocity from ROS in the topic cmd_vel, and then graph the angular and linear velocity from the gps, comparing the two graph it's a way to test the response of the robot to the requirements of the navigation stack

*Thread Reply:* i'm do that to test the response of my robot

*Thread Reply:* if you put the two graphs at the same axis (time) you will see a delay between the sent w and v, and the final w and v of the robot. If these delay are too high, the navigation stack could not work propely

*Thread Reply:* this delay can be reduced with acclimx: acclimtheta: from the teblocalplanner_params.yaml

*Thread Reply:* and depends of the configuration of your low level controller too

*Thread Reply:* the PID parameters

*Thread Reply:* I haven't compared it to any inputs yet. I have been driving this with the joystick. One of the next experiments may be to load the cmd_vel to angle converter: https://github.com/rst-tu-dortmund/teb_local_planner/blob/0d242c0aff4f81b4ff5d67d7e4f970a8f454ad4a/scripts/cmd_vel_to_ackermann_drive.py Then I can specify an input rotational velocity and scale the angle out so my vehicle turns at the same rate.