Abstract: This article presents a hierarchical control framework for quadrupedal locomotion that unifies the complementary strengths of model-based optimization and reinforcement learning. We develop ...