Max Robotics — Global Service Robot Directory + US Certification

About

Muzhi Han is a robotics researcher whose work bridges the critical gap between 3D scene understanding and autonomous robot manipulation. His research centers on reconstructing interactive, functional 3D environments and enabling robots to learn novel manipulation skills within them. Han’s major contributions include redefining scene reconstruction from an embodied agent’s perspective—moving beyond mere geometric accuracy to emphasize actionable, functional constraints that support robot autonomy. His highly cited work on panoptic mapping and CAD model alignment (30 citations) and part-level scene reconstruction (8 citations) has established new paradigms for how robots perceive and interact with their surroundings. Han has also pioneered methods for agent-agnostic skill learning (Ag2Manip, 12 citations) and closed-loop, open-vocabulary mobile manipulation using GPT-4V (10 citations), demonstrating how large language models can enable adaptive planning and reasoning in unstructured environments. His diffusion-based trajectory optimization for mobile manipulation (M² Diffuser, 9 citations) further showcases his innovative approach to coordinating navigation and manipulation. With over 100 total citations and a publication record spanning top venues from 2021 to 2025, Han is rapidly establishing himself as a leading voice in embodied AI and interactive scene understanding.