Template Model Inspired Task Space Learning for Robust Bipedal Locomotion, by Guillermo A. Castillo, Bowen Weng, Shunpeng Yang, Wei Zhang, and Ayonga Hereid, was accepted to IROS 2023.
The paper combines a high-level reinforcement-learning planner with a model-based controller, using a reduced-order walking model to design the policy’s observations and actions. IROS 2023 took place in Detroit, Michigan, from October 1 to 5, 2023.
Read the paper and demonstration or visit the proceedings record.