Train Custom Robot Dog with RL Locomotio

Omniverse Isaac Sim
1

Hello,

I am currently stugling implementing reinforcement learning on my custom robot (Botzo GitHub) using Isaac Lab and the provided example under /source/isaaclab_task/isaaclab_task/manager_based/locomotion.

First of all, here the URDF of my robot. Then I built the robot asset configuration (see here). After I just duplicate the config file to train A1 robot dog to learn stable walk. I adapt it to my Botzo (see and test here). But I am getting crazy results. The robot is not learning at all. The reward starts very low and grow arround zero, on the other hand the mean eppisode lenght time starts very high and decrease to neraly zero. The robot, just spawn and move the legs so crazy that it get reset immidiatly because it contact with the ground!

(env_isaaclab) PS C:\Users\grego\Desktop\GRINGO\IsaacLab\IsaacLab> python .\scripts\reinforcement_learning\rsl_rl\train.py --task=Isaac-Velocity-Flat-Botzo-v0 --headless --video

rl-video-step-134000:

Any suggestion would be very much appriciate to help me getting unstuck and actualy teach my robot how to walk using the predifine Isaac Lab scripts!

Thank you so much

2

Thank you for your interest in Isaac Lab. To ensure efficient support and collaboration, please submit your topic to its GitHub repo following the instructions provided on Isaac Lab’s Contributing Guidelines regarding discussions, submitting issues, feature requests, and contributing to the project.

We appreciate your understanding and look forward to assisting you.