The Dynamic Robotics and AI Lab (DRAIL) develops the algorithms and systems that enable robots—especially legged platforms such as humanoids and quadrupeds—to perform real-world work with reliability and agility. Our research spans the full autonomy stack, including whole-body control, deep learning, computer vision, reinforcement learning, and high-level planning, with the goal of allowing robots to both move and manipulate in complex environments.
DRAIL has been an early innovator in simulation-to-reality learning for humanoid locomotion, helping shape modern approaches to training robust controllers before deployment in the physical world. Building on this foundation, our current efforts focus on integrating advanced 3D perception, high-level task and motion planning, and whole-body manipulation and locomotion controllers to achieve next-generation autonomy for legged robots operating in real, unstructured environments.