My research centers around the motion synthesis (planning and control) for legged robots. I seek efficient, effective, and elegant algorithms that exploit the hybrid dynamics and feedback structure of underactuation to solve motion synthesis problems on physical robots through the use of a variety of mathematical techniques from dynamics modeling, control theory, numerical optimization, and machine learning. My ultimate goal is to develop principled and rigorous methodologies for realizing human-level physical intelligence on bipedal robotic systems that can move and work in our daily lives.