Legged Robotics

Legged robotics investigates how quadrupeds, humanoids, and other walking or hopping machines move safely and efficiently while interacting with uneven, cluttered, or changing terrain. Common approaches include model-predictive and whole-body control, reinforcement learning, foothold and terrain planning, and sensor fusion for estimating motion and foot contact. Researchers test these methods in simulation and on physical robots, with tasks ranging from rough-terrain traversal and jumping to manipulation and navigation. A central challenge is making robots robust to uncertain contact, sensor failures, and unfamiliar environments while balancing agility, energy use, safety, and reliable sim-to-real transfer.