Hierarchical jumping optimization for hydraulic biped wheel-legged robots

This paper addresses the challenging task of implementing dynamic jumping motions for biped wheel-legged robots, a motion mode that significantly enhances their mobility and adaptability to complex terrains. However, due to the under-actuation and dynamic instability nature of this type of robots, executing high-dynamic motions remains difficult. To tackle this issue, this paper proposes a novel Hierarchical Jumping Optimization (HJO) framework specifically designed for hydraulic biped wheel-legged robots. The framework involves an upper layer center of mass (CoM) trajectory optimization based on the dynamic properties of a dual-mass spring loaded inverted pendulum (DM-SLIP) model, along with the CoM dynamic coupling relationship introduced by the under-actuated nature of two-wheel self-balancing robots. This leads to a reduction in the dimensionality of optimization variables, facilitating effective CoM trajectory generation. Subsequently, the CoM motions of the robot in the Cartesian space are mapped into the joint space in real-time by using the middle layer hierarchical operational space control (HOSC). By incorporating centroid angular momentum as a task objective, the stability of the robot in highly dynamic motions can be enhanced. Finally, the proposed HJO framework is experimentally validated through jumping experiments on an actual robot. The experimental results demonstrate that the framework optimally utilizes the robot’s under-actuated characteristics, significantly reducing computational costs associated with planning and tracking, while simultaneously preserving motion accuracy and robust.