Papers
23
Total Citations
395
H-Index
11
About
Chuang Gan is a leading researcher at the intersection of embodied AI, robotics, and multimodal perception, with work spanning 3D scene understanding, physical simulation, and vision-language systems. His research addresses some of the most pressing challenges in building intelligent agents that can perceive, reason about, and act within complex physical environments. Among his most influential contributions is ConceptGraphs, a framework for constructing open-vocabulary 3D scene graphs that enable robots to efficiently perform perception and planning tasks — work that has already garnered 178 citations, reflecting its rapid adoption by the robotics and computer vision communities. His ThreeDWorld Transport Challenge established a rigorous benchmark for physically realistic embodied AI, pushing the field toward more grounded evaluation of task-and-motion planning. Through PlasticineLab and DiffSkill, Gan advanced differentiable physics simulation for deformable object manipulation, filling a critical gap in robotic learning environments. His more recent 3D-VLA work bridges 3D world modeling with vision-language-action generative frameworks, extending beyond the limitations of 2D-based approaches. Gan also explores multimodal integration, including asynchronous audio-visual reasoning. Collectively, his contributions represent a coherent, ambitious effort to ground AI agents in the rich physical and semantic complexity of the real world.
Research Focus
Key Achievements
Top Papers
- 1ConceptGraphs: Open-Vocabulary 3D Scene Graphs for Perception and Planning178 citations · 2024
- 2
- 3
- 4RoboNinja: Learning an Adaptive Cutting Policy for Multi-Material Objects16 citations · 2023
- 5
- 6
- 73D-VLA: A 3D Vision-Language-Action Generative World Model14 citations · 2024
- 8Finding Fallen Objects Via Asynchronous Audio-Visual Integration13 citations · 2022
- 9
- 10ConceptGraphs: Open-Vocabulary 3D Scene Graphs for Perception and Planning12 citations · 2023