Planning and Execution of Dynamic Whole-body Locomotion for a Wheeled Biped Robot on Uneven Terrain

To improve the adaptability of the wheeled biped robot (WBR) to uneven terrain, firstly an integrated modeling method for wheeled-legs is proposed. The under-actuated part is effectively restrained by defining the interaction force between the WBR and the trunk. The mapping relationship between the wheeled leg’s end force and the joint torques in the balanced state is built. Based on this premise, a control framework that does not rely on external sensors is proposed, and the trunk pose is used as the task space to plan the generalized force output of the wheeled legs and calculate the joint torques. Since the joint space position is not constrained, the leg wheels will be based on the terrain conditions and are adaptively stretched and adjusted back and forth. To further improve the terrain adaptability, a slope estimator and a stabilizer are constructed to deal with the attitude fluctuation caused by the sudden change of terrain. The control framework is proved to verify by simulations and experiment.