Due to the lack of documentation, or my inability to locate documentation related to the examples provided by the Isaac Sim team, such as the ant_instanceable robot;
I am currently engaged in reverse-engineering the steps to find answers to the following questions:
Why does the Issac Sim team use different Xforms in the ant_instanceable.usd model?
How can we make adjustments in various files (ant_env_cfg.py, yaml, train.py, etc.) after making changes in the ant_instanceable.usd model to enable training using RL_games on the ‘Isaac-Ant-v0’ environment?
Could you kindly provide complete documentation for writing different files (ant_env_cfg.py, yaml, train.py, etc.) from scratch? This is particularly needed if we want to train models other than the aforementioned one, making them capable of training using RL_games on the ‘Isaac-Ant-v0’ environment.
Please let me know if you need any further clarification.
Warmest regards,
Mohammad Shoaib
P.S. I’m using Issac sim 2023.1.1 w/ below PC specifications;
It goes over all the steps around creating an environment and training with Stable-Baselines3. The steps are the same for RL-Games or any other framework in Orbit.
Regarding tuning the parameters, you can do those in the config files inside every environment folder.