Nonlinear MPC-Based Control Framework for Quadruped Robots: Touch-Down in Complex Terrain

Quadruped robots show potential beyond that of wheeled robots for walking over complex terrain thanks to the bionic leg design, and to take advantage of this, we propose a control framework for a quadruped robot to achieve touch-down in a designated area, aiming to help the robot overcome the challenge in complex terrains. We incorporate a foot placement constraint into a nonlinear model predictive control (MPC) framework guided by global localization, enabling the robot to touch-down in the specified area while maintaining balance. Furthermore, a whole full-body controller is utilized to track the optimized trajectory generated by the MPC. Both simulation and hardware experiments demonstrate the effectiveness of our proposed control framework.