Locomotion Adaption for Hydraulic Humanoid Wheel-Legged Robots Over Rough Terrains

Abstract

Compared with the traditional hydraulic humanoid robots, the WLR-II, a novel hydraulic wheel-legged robot developed by using hose-less design, can significantly increase the reliability and maneuverability. The WLR-II combines the rough-terrain capability of legs with the efficiency of wheels. In this paper, a novel framework called rough-terrain adaption framework (RTAF) is presented which allows WLR-II to move on both flat terrains and terrains with unmodeled contact dynamics. RTAF is a hierarchical framework, which has a high-level balance controller and a low-level impedance controller that a high-performance nested torque controller with feed-forward velocity compensation is used. The low-level impedance controller for the hydraulic-driven unit can cancel out the load dynamics influence such as unexpected terrain disturbances and increase the force-tracking performance. With the high-level balance controller, the robot is able to handle unexpected terrain disturbances through wheel-ground force estimation, pitch/roll balance control and impedance parameter regulator. The proposed approach is suitable for a wheel-legged humanoid robot to manage balance through torque control at joints and regulate force-based interaction on rough terrains. The performance of the proposed RTAF is evaluated on variable gradient slopes and grassland which are the typical rough-terrain scenarios for real-world applications. The experimental results reveal that the maximum speed of grassland movement can reach 3 km/h.