Abstract:Hopping is a challenging action of bipedal robots and is critical in robotic motions such as obstacle crossing. However, the inherent dynamics of the robot are so complex that only Superior-performance planning and control algorithms can achieve perfect hopping. Unfortunately, nonlinear model predictive control (NMPC) isn’t easy to apply to a physical robot because it is computationally expensive. In this paper, we propose a planning method based on the spring loaded inverted pendulum (SLIP) model to generate the center of mass (CoM) trajectory for periodic hopping, and combine it with quadratic programming (QP)-based whole-body control (WBC) to resolve the control problem in real time. Since WBC can deal with the full dynamics of the robot and various constraints, SLIP model only needs to plan the robot CoM trajectory and the plantar force to provide a reference for the continuous hopping of the bipedal robot, thus greatly reducing the computational cost of the full dynamics model planning. The approach’s validity was verified in simulation on a heavy-legged bipedal robot. Furthermore, our method is more adaptive than traditional control methods for heavy-legged biped robots.

Hopping Motion on Heavy-Legged Bipedal Robot Based on SLIP Model and Whole Body Control