Template-based hopping control of a bio-inspired segmented robotic leg

Abstract

In human hopping in place, the axial leg function is representable by a spring mass model. This description can be utilized to control robot hopping. In this paper, the SLIP (spring loaded inverted pendulum) model is employed as a template for the control of MARCO Hopper II, a robot with a two-segmented leg. Using VMC (virtual model control) a spring is emulated between the foot and hip joint. The required knee torque is generated by a cable-driven actuator to mimic the unilateral knee extensor. In ground contact, gravity acts as the antagonistic knee flexor. The paper describes an evolution of controllers operating on systems ranging from a simple SLIP to more complex simulation models and finally proposes a control strategy that yields stable hopping in the hardware setup. To compensate losses, energy management by tuning the virtual leg spring stiffness is utilized. The resulting hopping motion is similar to human motions with respect to the positions of foot and hip as well as the ground reaction force. A combination of the SLIP model with a control technique for segmented structures and the addition of a bio-inspired energy management method is the result of this work.