Representation-Free Model Predictive Control for Dynamic Motions in Quadrupeds

This paper presents a novel Representation-Free Model Predictive Control (RF-MPC) framework for controlling various dynamic motions of quadrupedal robot in three dimensional (3D) space. Our formulation directly represents the rotational dynamics using rotation matrix, which liberates us from issues associated with use Euler angles and quaternion as orientation representations. With variation-based linearization scheme carefully constructed cost function, MPC control law is transcribed to standard Quadratic Program (QP) form. The controller can operate at real-time rates 250 Hz on quadruped robot. Experimental results including periodic gaits controlled backflip validate that our strategy could stabilize involve singularity 3D maneuvers.