John W. Roberts
Papers
1
Total Citations
21
H-Index
1
About
John W. Roberts is a leading researcher in robotics and machine learning, specializing in reinforcement learning under model uncertainty. His work addresses a fundamental challenge: enabling robots to act effectively in complex, poorly understood environments where true world dynamics are unknown. His most-cited paper, "Reinforcement learning with misspecified model classes" (2013, 21 citations), tackles this by analyzing how learners can compensate for unknown dynamics when provided with a class of models, typically selected via minimum prediction error metrics. This contribution has been pivotal in advancing robust decision-making for real-world robotic systems. Roberts’ research bridges theoretical guarantees and practical deployment, influencing fields like adaptive control and autonomous navigation. His work is widely recognized for its rigor and applicability, making him a key figure in the development of resilient AI systems. With a growing citation impact, Roberts continues to shape how robots learn and adapt in uncertain environments.
Research Focus
Key Achievements
Top Papers
- 1Reinforcement learning with misspecified model classes21 citations · 2013