Flip Stunts on Bicycle Robots Using Iterative Motion Imitation
Abstract¶
This work demonstrates a front-flip on bicycle robots via reinforcement learning, particularly by imitating reference motions that are infeasible and imperfect. To address this, we propose Iterative Motion Imitation (IMI), a method that iteratively imitates trajectories generated by prior policy rollouts. Starting from an initial reference that is kinematically or dynamically infeasible, IMI helps train policies that lead to feasible and agile behaviors. We demonstrate our method on Ultra-Mobility Vehicle (UMV), a bicycle robot that is designed to enable agile behaviors. From a self-colliding table-to-ground flip reference generated by a model-based controller, we are able to train policies that enable ground-to-ground and ground-to-table front-flips. We show that compared to a single-shot motion imitation, IMI results in policies with higher success rates and can transfer robustly to the real world. To our knowledge, this is the first unassisted acrobatic flip behavior on such a platform.
Summary¶
Introduces Iterative Motion Imitation for learning agile bicycle-robot flips from imperfect or dynamically infeasible reference motions. The paper demonstrates a practical loop where policy rollouts become improved references, enabling feasible stunt behaviors that direct imitation of the initial trajectory cannot produce.