Discrete-Time Hybrid Automata Learning: Legged Locomotion Meets Skateboarding

Authors: Hang Liu, Sangli Teng, Ben Liu, Wei Zhang, Maani Ghaffari
Website: https://umich-curly.github.io/DHAL/
Paper: https://arxiv.org/pdf/2503.01842
Contact: hangliu@umich.edu

Install

1. Create environment and install torch

   conda create -n dhal python=3.8 
   pip3 install torch torchvision torchaudio 
   

2. Install Isaac Gym preview 4 release https://developer.nvidia.com/isaac-gym

   unzip files to a folder, then install with pip:

   cd isaacgym/python && pip install -e .

3. Clone our repo and install

   git clone git@github.com:UMich-CURLY/DHAL.git
   cd DHAL/legged_gym
   pip install -e .
   cd ../rsl_rl
   pip install -e .
   

Training

• go to legged_gym/legged_gym/scripts

  python train.py --exptid=dhal
  

Play

• go to legged_gym/legged_gym

  python play.py --exptid=dhal
  

Arguments

• --exptid: string, can be WHATEVER and the weights would be saved in corresponding directory
• --checkpoint: the specific checkpoint you want to load. If not specified load the latest one.
• --resume: resume training from previous checkpoint, need to use with --exptid.
• --wandb: with wandb logging.
• --debug: debug mode, less agents.

Acknowledgement

Our code are based on these previous outstanding repo:

• https://github.com/leggedrobotics/rsl_rl
• https://github.com/leggedrobotics/legged_gym
• https://github.com/chengxuxin/extreme-parkour
• https://github.com/MarkFzp/Deep-Whole-Body-Control

Citation

If our work does help you, please consider citing us and the following works:

@misc{liu2025discretetimehybridautomatalearning,
      title={Discrete-Time Hybrid Automata Learning: Legged Locomotion Meets Skateboarding}, 
      author={Hang Liu and Sangli Teng and Ben Liu and Wei Zhang and Maani Ghaffari},
      year={2025},
      eprint={2503.01842},
      archivePrefix={arXiv},
      primaryClass={cs.RO},
      url={https://arxiv.org/abs/2503.01842}, 
}

We used codes in Legged Gym and RSL RL, based on the paper:

• Rudin, Nikita, et al. "Learning to walk in minutes using massively parallel deep reinforcement learning." CoRL 2022.