Papers
15
Total Citations
635
H-Index
7
About
Fangwei Zhong is a pioneering researcher at the intersection of computer vision, robotics, and reinforcement learning, with particular expertise in active object tracking, simultaneous localization and mapping (SLAM), and dexterous robotic manipulation. His most celebrated contribution, "Detect-SLAM" (2018), demonstrated how object detection and SLAM could be made mutually reinforcing rather than treated as isolated tasks, earning over 300 citations and establishing him as a leading voice in intelligent robotic perception. Building on this foundation, Zhong advanced the field of active object tracking through end-to-end reinforcement learning frameworks, developing systems capable of real-world deployment — work that has collectively attracted nearly 200 citations. His research consistently bridges simulation and physical deployment, as evidenced by CRAVES, which enables cost-effective robotic arm control through vision alone. More recently, Zhong has pushed boundaries in bimanual dexterous manipulation with Bi-DexHands and generalized embodied navigation, tackling some of robotics' most formidable open challenges. Spanning perception, control, and autonomous decision-making, his body of work represents a coherent and ambitious effort to build robots that see, reason, and act with human-like adaptability.
Research Focus
Key Achievements
Top Papers
- 1Detect-SLAM: Making Object Detection and SLAM Mutually Beneficial304 citations · 2018
- 2
- 3CRAVES: Controlling Robotic Arm With a Vision-Based Economic System55 citations · 2019
- 4Pose-Assisted Multi-Camera Collaboration for Active Object Tracking42 citations · 2020
- 5Bi-DexHands: Towards Human-Level Bimanual Dexterous Manipulation39 citations · 2023
- 6
- 7
- 8
- 9GraspARL: Dynamic Grasping via Adversarial Reinforcement Learning5 citations · 2022
- 10