A bi-level nonlinear predictive control scheme for hopping robots with hip and tail actuation

Abstract

A control concept is presented for hopping robots with hip and tail actuation. The flight phase is controlled by a novel nonlinear control concept that accounts for state and input contraints on the hip and tail while pursuing a linear error dynamics for the desired landing angle of the leg. An additional nonlinear model predictive control (MPC) scheme is superposed to coordinate the hopping cycles and to maximize the hopping speed. The MPC can be designed with a simple nonlinear optimization algorithm, as the constraints are already accounted for by the cascaded controller. Simulation results for a nonlinear dynamical model of the Festo BionicKangaroo show the working principle of the bi-level predictive control scheme.