Recent Publications: 'A Careful Examination of Large Behavior Models for Multitask Dexterous Manipulation' (2025); 'CodeDiffuser: Attention-Enhanced Diffusion Policy via VLM-Generated Code for Instruction Ambiguity' (2025); 'OmniShape: Zero-Shot Multi-Hypothesis Shape and Pose Estimation in the Real World' (2025).
Research Experience
Currently a Senior Research Scientist in the Large Behavior Models research team at Toyota Research Institute. Recent work includes building zero-shot algorithms to support robotic manipulation and investigating foundation models for vision and robotics.
Education
Ph.D. in Aeronautics and Astronautics from MIT, advised by Dr. Nicholas Roy; B.S. in Mechanical Engineering from the University of California, San Diego.
Background
Research Interests: Robot Learning, Object-Centric Representations, Zero-Shot Vision, Embodied Foundation Models. Background: I am a researcher working at the intersection of robotics, machine learning, and computer vision to equip embodied agents with greater understanding of the world for better decision making. I'm particularly excited about enabling general-purpose embodied agents by bridging 3D understanding and embodied actions in the real world to reconstruct, reason, and react.
Miscellany
Contact: You can reach me via email at katherine.liu@gmail.com.