I have developed a 6 dof bipedal robot that walks in omniverse using reinforcement learning.
I want to record the joint positions for 1 of the robots over time and save it to a file so I can test it on the actual robot without having to run inference, or test it in a different simulation environment so can assess for robustness, or even fit the joint position-time data to a curve so can compress the data so it can be output non inference endpoint such as an arduino
How do i extract the joint position data over time efficiently or over GPU? either during training or during inference i.e, test=True?
Hi there, you can use the get_joint_positions() API in ArticulationView to extract joint positions (Core [omni.isaac.core] — isaac_sim 2023.1.0-rc.36 documentation (nvidia.com)). This API can be called at each frame or at the desired interval to retrieve the joint positions of the robots and saved in a list, which can then be saved out to a file through pytorch or numpy.
thanks @kellyg I think the problem for me is that there is no easy way of creating a UI in omniverse to trigger the data collection and select the instance of the robot I want. Suspect this is something that will eventually need to be handled as a feature in Isaac Sim itself for doing asynchronous activity but I’ve decided to go for an approach that uses sockets for my rl python program which uses the IsaacSIM version of python to communicate with a tkinter UI running a native version of python. That way I can have 2 way communication and save out/visualise/process the data however I like without being restricted by the library limitations of the IsaacSIM library availability.
this is a little similar to webots methods I have used previously in other projects - I’ll try and post a gif once I have it working - I suspect I will show the robot instance I’m using by coloring it differently
