Mingtao Feng
Papers
15
Total Citations
295
H-Index
9
About
Mingtao Feng is a leading researcher at the intersection of robotics, computer vision, and artificial intelligence, with a primary focus on multi-robot navigation and 3D scene understanding. His most impactful work, "Transformer-Based Imitative Reinforcement Learning for Multirobot Path Planning" (103 citations), introduces a novel decentralized approach that enables large-scale robot teams to generate efficient, collision-free paths through imitative learning, addressing a critical bottleneck in scalable multi-agent systems. Feng has also made pioneering contributions to 3D scene graph prediction, developing hyperrectangle embedding and history-enhanced reasoning methods that allow autonomous robots to build rich, high-level environmental representations from RGB-D sequences—work recognized with 41 and 28 citations, respectively. His research extends to safe distributed navigation using variational Bayesian models (22 citations) and industrial applications, including robust pixel-wise prediction for robotic grasping and 3D reconstruction of profiled blades. Feng’s work on spatial-temporal networks for multi-robot collision avoidance and fast-response gesture recognition for human-robot interaction further demonstrates his commitment to creating safer, more responsive autonomous systems. With over 270 total citations across his top papers, Feng is shaping the future of intelligent robotics.
Research Focus
Key Achievements
Top Papers
- 1Transformer-Based Imitative Reinforcement Learning for Multirobot Path Planning103 citations · 2023
- 2
- 3History-Enhanced 3D Scene Graph Reasoning From RGB-D Sequences28 citations · 2025
- 4
- 5
- 6
- 7
- 8
- 9STR: Spatial-Temporal RetNet for Distributed Multi-Robot Navigation10 citations · 2025
- 10