Robust bipedal walking with variable leg stiffness

The bipedal spring-mass model embodies important characteristics of human walking, and therefore serves as an important starting point in studying human-like walking for robots. In this paper, we propose to extend the bipedal spring-mass model with variable leg stiffness and exploit the potential of this model in order to mimic the human capability to continuously adapt the leg stiffness to different gaits and to overcome disturbances. In particular, we present a control strategy that uses the variable leg stiffness to stabilize the walker to a desired gait, with minimal influence on the natural gait dynamics. Using numerical simulations, it is shown that the proposed control strategy significantly improves the robustness of the walker to external disturbances.