Rishabh Jangir

Papers

Total Citations

116

H-Index

About

Rishabh Jangir is a researcher at the forefront of reinforcement learning (RL) and robotic manipulation, with a focus on enabling robots to generalize and adapt to real-world environments. His key research areas include self-supervised learning, policy adaptation, inverse reinforcement learning (IRL), and multi-view robotic control. Jangir’s most influential contribution is his work on "Self-Supervised Policy Adaptation during Deployment," which addresses the critical challenge of deploying RL-trained policies in unseen environments—a common hurdle in robotics. This work, with 58 citations, proposes a method that allows robots to continue learning and adapting after deployment without human supervision, significantly improving robustness in dynamic settings. He also advanced precision-based manipulation in "Look Closer: Bridging Egocentric and Third-Person Views With Transformers for Robotic Manipulation" (47 citations), where he introduced a transformer-based architecture that fuses egocentric and third-person visual perspectives, enabling fine-grained motor control from visual feedback alone. Additionally, his research on "Graph Inverse Reinforcement Learning from Diverse Videos" explores learning reward functions from varied video demonstrations, reducing the need for manual reward design. Jangir’s work is notable for its practical impact, bridging the gap between simulation and real-world deployment, and his innovative use of transformers and self-supervision positions him as a rising leader in robot learning.