Error in robot localization

[EmimaJiva]

@ggari-robotnik @sandra-Robotnik The docker image rb-theron-gazebo-noetic-0.1.0 uploaded in the repository dockerhub has been used. When the robots move, they lose their location, as can be seen in the following image.

[sandra-Robotnik]

@EmimaJiva, Is the problem when they move and/or when they stay idle for some time? Can you please record a video?

[EmimaJiva]

@ggari-robotnik @sandra-Robotnik The problem is both if they move and if they don't move. Once they are lost, they are no longer well located.

https://github.com/RobotnikIstobalAI2/rb_theron_sim/assets/121794419/b1d9fdc9-8fcc-4e4a-a66a-ee7231ddc13f

[sandra-Robotnik]

@EmimaJiva , Our experts require that you provide us with a video of the moment when the robots get lost so that they can analyze it. Can you record a video which shows this?

[sandra-Robotnik]

@EmimaJiva , and can you record a rosbag of the moment when the robots get lost?

[EmimaJiva]

@sandra-Robotnik The file of the recording is large and I cannot upload it here, I will send it to you by upv exchange.

[EmimaJiva]

@sandra-Robotnik Sorry, I sent you a bad file. Now I send you the good one.

The moment when the robot is lost is not always the same, sometimes it is lost early, sometimes it takes longer to get lost. But it always ends up getting lost.

[EmimaJiva]

@sandra-Robotnik Did you receive the correct file?

[ggari-robotnik]

@EmimaJiva we just sent you a Google drive link, were you can place the rosbag there. We have problems with microsoft platforms. You should have received a notification email on your account

[EmimaJiva]

@ggari-robotnik @sandra-Robotnik I have added another video and the rosbag in the Drive.

[sandra-Robotnik]

@EmimaJiva Our experts suspect that it could be related to a physical drift that it is pretty common on some robots in Gazebo. This drift can happen with the robot moving or idle.

For our experts it would be easier to check what is happening by seeing the simulation. So, I will send you an email to arrange a meeting.

[EmimaJiva]

@sandra-Robotnik Okay

[EmimaJiva]

@ggari-robotnik @sandra-Robotnik As your expert told me, I try to test different commits from the rb_theron_common repository of RobotnikAutomation. For this I modify the rb_theron_sim.repos file and put the commit as you can see in the figure. But when I build the new image, I get the following error: Can you tell me how to fix it? I will try to fix it but if you have the answer it would be faster.

[ggari-robotnik]

Try changing this here: rb_theron_sim.repos.yaml

  src/rb_theron_common:
    type: git
    # url: https://github.com/RobotnikAutomation/rb_theron_common
    url: https://github.com/RobotnikIstobalAI2/rb_theron_common.git
    version: noetic-devel

or with explicit commit:

  src/rb_theron_common:
    type: git
    # url: https://github.com/RobotnikAutomation/rb_theron_common
    url: https://github.com/RobotnikIstobalAI2/rb_theron_common.git
    version: e45f9884f6f0c6eb61063f3741c154d2fd9b7eee

[EmimaJiva]

@ggari-robotnik @sandra-Robotnik Trying it as you say, with an explicit commit and I get the same error. rb_theron_navigation_private does not exist in that commit, it is a folder that you have created.

[ggari-robotnik]

I'm puzzled, I'm looking on this repo on that commit and I see the folder rb_theron_navigation_private. Quite strange

@EmimaJiva Can you check that in the docker building process is taking the right commit and the folder where is downloaded has not changed?

To do that, add an ls command of rb_theron_common just before the catkin ignore and build it with the following directive in order to see better what it happening:

BUILDKIT_PROGRESS=plain docker compose build

[EmimaJiva]

@ggari-robotnik @sandra-Robotnik The rb_theron_common commit I am trying to test is 1e080cf1ea8f0f724f3371d265cd4ec4ebc38d30. I don't see the rb_theron_navigation_complete folder anywhere. Docker downloads ok that commit but it doesn't find that folder. Can you send me the link to where the folder is in that commit, please?

[ggari-robotnik]

@EmimaJiva can you try this commit e45f9884f6f0c6eb61063f3741c154d2fd9b7eee for rb_theron_common?

[EmimaJiva]

@ggari-robotnik I try this commit and when I launch it I obtained this error:

[EmimaJiva]

@ggari-robotnik @sandra-Robotnik In case the information helps you. The robot usually gets lost when it turns on itself and makes abrupt changes of orientation.

[sandra-Robotnik]

@EmimaJiva , Our experts have identified the origin of the problem in the simulation. It lies in the fact that the robot wheels are not actually stationary; instead, they are in constant motion. To solve it, it is necessary to add one parameter (one for each wheel) in the launch that spawns the model of the robot (rb_theron_robot.launch).

They will resolve it. However, the steps they will follow are as follows:

1. Add in the rb_theron_sim repository inside the repos folder in repos.yaml the RobotnikAutomation/gazebo_ros_pkgs repository.

• It takes 1 hour to compile.

2. Add the missing parameter (one for each wheel) in rb_theron_robot.launch located within rb_theron_sim > rb_theron_gazebo > launch before the (currently line 41).

• To setup the name of the parameter and its configuration, it is necessary to check the code of the lines from 220 to 248 of the default_robot_hw_sim.cpp file from the repository gazebo_ros_pkgs within the branch noetic-devel.

3. Configure the proper name of the parameter and select its value as position instead of velocity which is its default name.
   • As this parameter is a rosparam, it is possible to do a rosparam list to check its name
   • The name would be something like this: robot_namespace + "/gazebo_ros_control/pid_gains/" + joint_names_[j] + "/close_loop"

[sandra-Robotnik]

@EmimaJiva , I merged in rb_theron_sim in the branch noetic-devel the required changes to solve the wheels of the robot. Please, try the simulation using the noetic-devel branch to check that works fine. If the issue is solve, I will close this issue.

[EmimaJiva]

@ggari-robotnik @sandra-Robotnik I have tested the updated simulation and it works fine. There are some times that it gets a little lost (attached image) but it recovers its location without problem.

There is also an error related to the camera when launching with open-rmf-deployment (attached image), but it does not seem to affect the performance of the simulation.