search

ntnu-arl / aerial_gym_simulator

Aerial Gym Simulator - Isaac Gym Simulator for Aerial Robots
https://ntnu-arl.github.io/aerial_gym_simulator/
BSD 3-Clause "New" or "Revised" License
330 stars 57 forks source link

Load policy and run inference #6

Open mohamedamr13 opened 6 months ago

mohamedamr13 commented 6 months ago

Is there a script to load a train network ( rl_games ) and run inference similar to the play script but with the inferred actions instead of the constant actions in the play script.

mihirk284 commented 3 months ago

Hello, we have released a new version of the simulator last week. More features are added and better packaging with RL frameworks is provided. Please refer to this link to see how you can perform inference with the rl_games framework with the current version: https://ntnu-arl.github.io/aerial_gym_simulator/6_rl_training/#train-your-own-navigation-policies

The current package has an rl_training folder that provides scripts to train and test networks for control and navigation.

### Train the navigation policy for the quadrotor with velocity control
python3 runner.py --file=./ppo_aerial_quad_navigation.yaml --num_envs=1024 --headless=True

### Replay the trained policy in simulation
python3 runner.py --file=./ppo_aerial_quad_navigation.yaml --num_envs=16 --play --checkpoint=./runs/<weights_file_path>/<weights_filename>.pth