I am currently an MSR student at the Robotics Institute, Carnegie Mellon University, advised by Guanya Shi and Max Simchowitz.
I received my B.Eng. in Automation from Shanghai Jiao Tong University, advised by Yue Gao, Yeqiang Qian, and Ming Yang. In 2024 I was a visiting student at Cornell University, advised by Tapomayukh Bhattacharjee.
I work toward general-purpose robots that combine broad learned priors with grounded models of themselves and the physical world, enabling reliable and versatile behavior across tasks, embodiments, and environments.
Under review
TL;DR: A three-stage Planner-IDM framework for humanoid control: freezing the planner and finetuning only the inverse dynamics model on about two minutes of target-domain rollouts aligns it to new terrain, payload, or actuator dynamics — no rewards or demonstrations needed.
RSS 2026
TL;DR: Predicts how whole-arm contacts and forces will evolve by coupling learned latent dynamics with analytical kinematics through contact Jacobians, then plans with contact-aware MPC that trades task progress against force regulation.
RSS 2026
TL;DR: An open-source Python framework that lets robots, sensors, teleoperation interfaces, and policies be swapped independently, validated on VLA deployment across three morphologies and four hardware platforms.
CoRL 2025
TL;DR: When multiple contacts impose conflicting force preferences no controller can satisfy all of them, so PrioriTouch learns to rank the competing objectives and executes them through hierarchical operational space control.
IROS 2024
TL;DR: Finds non-repetitive LiDAR the more promising scanning pattern, on the strength of its superior object scanning, using Repetitive-or-not — a CARLA dataset that captures both patterns simultaneously and exposes the domain gap between them.
