Projects
Reinforcement Learning for Humanoid Robots
We apply the Learn to Teach framework to the Digit humanoid robot, targeting sample-efficient sim-to-real transfer for robust locomotion across diverse terrains and conditions. Our approach uses inverse reinforcement learning to learn reward functions from expert demonstrations, then distills a LiDAR-enriched teacher policy into a deployable proprioceptive student.
Distributional Inverse Reinforcement Learning
We introduce a distributional framework for offline inverse reinforcement learning that models uncertainty over reward functions and the full distribution of returns, rather than estimating a single reward. Using first-order stochastic dominance and distortion risk measures, the method captures richer structure in expert behavior, enabling risk-aware imitation and state-of-the-art performance across synthetic, neurobehavioral, and MuJoCo benchmarks.