Foot-end contact state estimation for quadruped robots based on proprioceptive sensing

Abstract To solve the problem of estimating the early or delayed foot-end contact state that may occur on uneven road surfaces for quadruped robots, a generalized momentum end-of-foot contact force estimation method combined with Kalman filtering without relying on external sensors is proposed, and an end-of-foot contact state estimation is realized based on Proprioceptive Sensing. First, the quadruped robot model structure and parameters are introduced. Then, based on the dynamics model and generalized momentum theory, the estimation of foot-end contact force is achieved by combining Kalman filtering, and the foot-end contact force estimation method is improved. Finally, the contact probability models of phase progress planning, external force at the foot end and height of the foot end from the ground are established, and the Kalman filter algorithm is used to fuse the above probability models to achieve the estimation of the contact state of the foot end of the quadruped robot. The experimental results show that the method can effectively improve the accuracy and robustness of the foot-end contact state estimation.