Hierarchical optimum control of a novel wheel-legged quadruped

This paper presents an optimal control architecture for Pegasus, a novel quadruped wheel-legged robot with hybrid locomotion capabilities. The proposed control architecture comprises of a hierarchical motion planner and a model predictive controller (MPC) that optimizes motion planning and control in various stages. A command-based motion planner is implemented to map desired robot states to optimal joint positions and velocities. This enables the MPC to seamlessly integrate legged and wheeled locomotion as a single task. The legs are modeled as N-link manipulators, and parallel tracking MPC controllers are implemented to optimize torques. This approach results in improved motion control and comprehensive four-wheel independent steering mechanism maneuvers. The experiments and results demonstrate the practical feasibility and robustness of the proposed control approach, with Pegasus exhibiting stable balancing, precise motion control, and the ability to navigate through challenging paths. Overall, the proposed control architecture provides a promising solution for achieving hybrid locomotion capabilities in quadruped wheel-legged robots.