Gokul Narayanan
Papers
2
Total Citations
18
H-Index
2
About
Gokul Narayanan is a robotics researcher whose work bridges the gap between learned policies and real-world industrial dexterity. His primary research areas include robot manipulation, haptic sensing, and vision-based learning for manufacturing automation. Narayanan’s most impactful contribution, “Learning on the Job,” introduces a self-rewarding offline-to-online finetuning framework that enables robots to adapt to novel industrial insertion tasks—such as handling unfamiliar connectors—without requiring new training data. This work, with 13 citations, addresses a critical limitation in robotics: the failure of learned policies to generalize to unseen scenarios. By leveraging a reward classifier that remains robust even when the policy falters, his method allows robots to “learn on the job” in real-world settings. In earlier work, Narayanan explored haptic object parameter estimation during within-hand manipulation, demonstrating that robots can estimate object properties through touch alone—a vital capability for environments where vision fails due to occlusion or poor lighting. His research has direct implications for manufacturing, where flexible automation remains a grand challenge. Narayanan’s work exemplifies how combining self-supervised learning with tactile feedback can push robots toward greater autonomy and adaptability in industrial applications.
Research Focus
Key Achievements
Top Papers
- 1
- 2