David Held

David Held is an Associate Professor in the Robotics Institute at Carnegie Mellon University, where he leads research at the intersection of robotics, machine learning, and computer vision. His work focuses on enabling robots to perceive, predict, and act in complex, cluttered, and dynamic real-world environments. Held develops deep learning and probabilistic modeling approaches that help robots understand how objects move, how environments change, and how robotic actions can influence the world to achieve specific goals. His group pursues applications in both robotic manipulation and autonomous driving, addressing challenges in segmentation, tracking, velocity estimation, multi-task learning, and robust control. Current projects explore simulation-to-real transfer, safe learning-based control policies, and methods that generalize across diverse manipulation tasks. Ultimately, his research aims to expand the capabilities of robots operating in everyday human environments—from homes and assisted living facilities to hospitals, schools, and disaster-response scenarios—through a principled integration of perception and action.