Dynamics Modeling and Control Architecture for Efficient, Manoeuvrable and Robust Monoped Hopping over Rough Terrain

Abstract

Leg dynamics and control have been widely studied using mass-spring systems such as the Spring Loaded Inverted Pendulum (SLIP) model [1]. The SLIP model is commonly accepted as the simplest model that resembles leg dynamics. While simplicity facilitates the understanding of the basic aspects of system behavior, it can result in neglecting some essential factors from control perspective such as leg mass, hip actuation and energy dissipation. Neglecting such aspects increases the gap between simulated results and what can be achieved in reality. Furthermore it makes the transfer of the gained controllers onto robotic hardware difficult. We thus aim at preparing sufficiently detailed mathematical model that can accurately describe the dynamical properties of our monoped hopping robot. We then exploit this detailed knowledge to design control laws which provide a more energy efficient, manoeuvrable and robust behavior in unstructured environment. We validate the proposed method in a challenging scenario where the hopper is expected to pass through rough/sloped terrain while maintaing a desired speed and acting in the vicinity of an energy-optimized gait.