Optimal Control

Optimal control formulations and solvers for safe, high-performance decision-making and trajectory generation in dynamic systems.

Learning from Demonstrations

Learning control policies from expert demonstrations through imitation learning and inverse reinforcement learning.

Reinforcement Learning from Human Feedback in Control Systems

Incorporating human feedback into reinforcement learning to align controller behavior with human preferences and intent.

Vision-Language Models for Autonomous Driving

Applying vision-language (VLM) and vision-language-action (VLA) models to autonomous driving — for scene understanding, reasoning, and language-conditioned decision-making and control.

Data-efficient Robot Learning

We exploit the geometric structure of robotics tasks — symmetry and equivariance (e.g. SE(3)) — so that models learn far more data-efficiently and generalize from only a handful of demonstrations.

Robot Foundation Models

We leverage foundation models pre-trained on large-scale vision, language, and video data to build Vision-Language-Action (VLA) and World-Action Models (WAM) for robots.

Task Planning with Prior Knowledge

We make robot task planning more efficient by injecting prior knowledge (e.g. commonsense knowledge) represented with structures such as ontologies — so robots reason with far less search and supervision.