Kefan Jin

Papers

Total Citations

H-Index

About

Kefan Jin is a pioneering researcher in autonomous multirobot systems and robust navigation, whose work bridges the gap between perception and motion control in complex, real-world environments. His primary research areas include visuomotor reinforcement learning, cooperative multirobot coordination, and resilient trajectory planning under uncertainty. Jin’s most influential contribution is a fully end-to-end learning framework for multirobot cooperative navigation, which integrates graph neural networks with deep reinforcement learning to enable local motion coordination from raw visual observations—a breakthrough cited 37 times. He further advanced the field with his Timed-Elastic-Band-Based Variable Splitting (TEB-VS) framework, which addresses critical stability issues in autonomous trajectory planning by incorporating constraint-based variable splitting. Jin has also tackled the challenge of environmental disturbances in self-driving applications, proposing a relation learning method for information reliability representation that enhances navigation robustness against sensor noise, lighting variations, and adversarial perturbations. His work is notable for its practical orientation, directly addressing the ghosting and instability problems that plague real-world autonomous systems, making him a key figure in developing more reliable and cooperative robotic teams.