Installation

1. Create a new python virtual env with python 3.6, 3.7 or 3.8 (3.8 recommended)
2. Install pytorch 1.10 with cuda-11.3:
   • pip3 install torch==1.10.0+cu113 torchvision==0.11.1+cu113 torchaudio==0.10.0+cu113 -f https://download.pytorch.org/whl/cu113/torch_stable.html
3. Install Isaac Gym
   • Download and install Isaac Gym Preview 3 (Preview 2 will not work!) from https://developer.nvidia.com/isaac-gym
   • cd isaacgym/python && pip install -e .
   • Try running an example cd examples && python 1080_balls_of_solitude.py
   • For troubleshooting check docs isaacgym/docs/index.html)

CODE STRUCTURE

1. Each environment is defined in an config file ($ROBOT_$TERRAIN_config.py) in configs, such as d1_flat_config.py. Task d1_flat_config.py as an example. The config file contains:
   • D1Flat class: containing reward functions and env settings.
   • D1FlatCfg class: containing all the environment parameters for training.
   • D1FlatCfg_Play class: containing all the environment parameters for playing.
   • D1FlatCfgPPO class: containing all the training parameters.
2. Both env and config classes use inheritance.
3. Each non-zero reward scale specified in cfg will add a function with a corresponding name to the list of elements which will be summed to get the total reward.
4. Tasks must be registered using task_registry.register(name, EnvClass, EnvConfig, TrainConfig). This is done in $ROBOT/__init__.py, but can also be done from outside of this repository.

Usage

1. Train:
   python scripts/train.py --task=d1_flat
   • To run on CPU add following arguments: --sim_device=cpu, --rl_device=cpu (sim on CPU and rl on GPU is possible).
   • To run headless (no rendering) add --headless.
   • Important: To improve performance, once the training starts press v to stop the rendering. You can then enable it later to check the progress.
   • The trained policy is saved in ./logs/<experiment_name>/<date_time>_<run_name>/model_<iteration>.pt. Where <experiment_name> and <run_name> are defined in the train config.
   • The following command line arguments override the values set in the config files:
   • --task TASK: Task name.
   • --resume: Resume training from a checkpoint
   • --experiment_name EXPERIMENT_NAME: Name of the experiment to run or load.
   • --run_name RUN_NAME: Name of the run.
   • --load_run LOAD_RUN: Name of the run to load when resume=True. If -1: will load the last run.
   • --checkpoint CHECKPOINT: Saved model checkpoint number. If -1: will load the last checkpoint.
   • --num_envs NUM_ENVS: Number of environments to create.
   • --seed SEED: Random seed.
   • --max_iterations MAX_ITERATIONS: Maximum number of training iterations.
2. Play a trained policy:
   python scripts/simple_play.py --task=d1_flat_play
   • By default, the loaded policy is the last model of the last run of the experiment folder.
   • Other runs/model iteration can be selected by setting load_run and checkpoint in the train config.
   • Export policy file is saved in ./model.pt and ./policy.onnx

Troubleshooting

1. If you get the following error: ImportError: libpython3.8m.so.1.0: cannot open shared object file: No such file or directory, do: sudo apt install libpython3.8. It is also possible that you need to do export LD_LIBRARY_PATH=/path/to/libpython/directory / export LD_LIBRARY_PATH=/path/to/conda/envs/your_env/lib(for conda user. Replace /path/to/ to the corresponding path.).

Known Issues

1. The contact forces reported by net_contact_force_tensor are unreliable when simulating on GPU with a triangle mesh terrain. A workaround is to use force sensors, but the force are propagated through the sensors of consecutive bodies resulting in an undesirable behaviour. However, for a legged robot it is possible to add sensors to the feet/end effector only and get the expected results. When using the force sensors make sure to exclude gravity from the reported forces with sensor_options.enable_forward_dynamics_forces. Example:

    sensor_pose = gymapi.Transform()
    for name in feet_names:
        sensor_options = gymapi.ForceSensorProperties()
        sensor_options.enable_forward_dynamics_forces = False # for example gravity
        sensor_options.enable_constraint_solver_forces = True # for example contacts
        sensor_options.use_world_frame = True # report forces in world frame (easier to get vertical components)
        index = self.gym.find_asset_rigid_body_index(robot_asset, name)
        self.gym.create_asset_force_sensor(robot_asset, index, sensor_pose, sensor_options)
    (...)

    sensor_tensor = self.gym.acquire_force_sensor_tensor(self.sim)
    self.gym.refresh_force_sensor_tensor(self.sim)
    force_sensor_readings = gymtorch.wrap_tensor(sensor_tensor)
    self.sensor_forces = force_sensor_readings.view(self.num_envs, 4, 6)[..., :3]
    (...)

    self.gym.refresh_force_sensor_tensor(self.sim)
    contact = self.sensor_forces[:, :, 2] > 1.

Acknowledgment

The project uses some code from the following open-source code repositories:

• legged_gym
• LocomotionWithNP3O

Any Questions?

If you have any more questions, please create an issue in this repository.