Nan Zeng

Chinese Academy of Sciences, Northeastern University

Papers

Total Citations

H-Index

About

Nan Zeng’s research lies at the intersection of robotics, computer vision, and neural computation, with a focus on enabling machines to perceive and act in complex environments. In his foundational 1993 work, “A Neural Network Based Inverse Kinematics Solution in Robotics,” Zeng pioneered the use of neural networks to solve the notoriously difficult inverse kinematics problem—a critical step for precise robotic arm control. This early contribution, cited 12 times, demonstrated how learning-based approaches could replace traditional analytical methods. A decade later, Zeng advanced robot perception with his 2002 paper on categorical color projection for road following. Building on Yue Du’s 1994 categorical color theory, he showed that representing color space categorically—rather than with raw RGB—significantly improved robot vision performance in challenging outdoor tasks. Using the SCARF system as a testbed, Zeng’s comparative experiments revealed that categorical color algorithms outperformed both RGB and intensity-based methods on difficult road-following scenarios. This work, with 10 citations, highlighted the practical value of biologically-inspired color representations for autonomous navigation. Zeng’s career exemplifies how bridging neural computation and vision can yield elegant, efficient solutions for real-world robotics challenges.