John Weng
Papers
4
Total Citations
146
H-Index
3
About
John Weng is a pioneering researcher in developmental robotics and autonomous learning systems, whose work bridges artificial intelligence, computer vision, and cognitive science. His primary research areas include developmental robotics, incremental learning, and biologically inspired vision systems for robot navigation and manipulation. Weng's most influential contribution is his work on value systems for developmental robots, which models how robots can learn from novelty and reinforcement—a framework that has garnered 119 citations and fundamentally shaped how robots acquire skills autonomously over time. He introduced the concept of a low-level value system that signals salient sensory inputs and modulates action mappings, enabling robots to develop behaviors without explicit programming. Weng also developed the SHOSLIF framework for vision-guided robot manipulation, allowing hand-eye systems to learn and recall action sequences from training examples. His work on incremental learning for vision-based navigation, using hierarchical recursive partition trees, demonstrated how robots can autonomously navigate by retrieving learned experiences. Additionally, Weng's Staggered Hierarchical Mapping (SHM) model, inspired by human early visual pathways, advanced outdoor navigation for developmental robots like SAIL. Through these contributions, Weng has established himself as a foundational figure in creating robots that learn and develop like biological systems.
Research Focus
Key Achievements
Top Papers
- 1Novelty and Reinforcement Learning in the Value System of Developmental Robots119 citations · 2002
- 2Incremental learning for vision-based navigation16 citations · 1996
- 3
- 4