Thai Duong
Papers
14
Total Citations
143
H-Index
7
About
Thai Duong is a robotics researcher whose work sits at the intersection of physics-informed machine learning, robot dynamics, and autonomous systems. He is best known for pioneering the application of Hamiltonian mechanics to neural network architectures, particularly through his development of Hamiltonian-based Neural ODE Networks on the SE(3) manifold — a framework that enables robots to learn accurate, physically consistent dynamics models directly from data. This foundational contribution, along with its port-Hamiltonian extension to Lie groups, has collectively garnered over 50 citations and established Duong as a leading voice in structure-preserving learning for robotics. His research spans adaptive control under disturbance, occupancy mapping for autonomous navigation, and multi-robot coordination, where his LEMURS algorithm and physics-informed multi-agent reinforcement learning approaches demonstrate a consistent commitment to scalable, stable, and generalizable robot autonomy. More recently, Duong has expanded into task planning with large language models, reflecting a broad vision for integrating semantic reasoning with rigorous physical modeling. Across his growing publication record, Duong's work is distinguished by its principled fusion of geometric mechanics and modern machine learning — producing controllers that are not only data-efficient but provably stable, making his contributions particularly valuable for safe deployment of robots in real-world, dynamic environments.
Research Focus
Key Achievements
Top Papers
- 1
- 2
- 3Optimal Scene Graph Planning with Large Language Model Guidance17 citations · 2024
- 4
- 5
- 6
- 7LEMURS: Learning Distributed Multi-Robot Interactions8 citations · 2023
- 8
- 9
- 10Optimal Scene Graph Planning with Large Language Model Guidance4 citations · 2023