Ujjwal Puri

Papers

Total Citations

H-Index

About

Ujjwal Puri is a robotics researcher focused on enabling general-purpose machines through advanced machine learning techniques. His work centers on multi-task and continual learning for robotic manipulation, addressing the critical challenge of how robots can not only adapt existing skills but acquire entirely new ones in real-world environments. In his 2022 paper "Efficient Multi-Task Learning via Iterated Single-Task Transfer" (4 citations), Puri introduced a novel approach that streamlines the learning of multiple manipulation tasks by iteratively transferring knowledge from single-task solutions, reducing computational overhead while maintaining performance. His earlier 2021 work, "A Simple Approach to Continual Learning by Transferring Skill Parameters" (2 citations), laid the groundwork for this paradigm, demonstrating how robots can build upon previously learned skills without catastrophic forgetting. Though his citation counts are modest, reflecting the early stage of his career, Puri’s contributions are technically significant, offering practical pathways toward robots that can operate as effective general-purpose machines—a holy grail in robotics. His research sits at the intersection of reinforcement learning, transfer learning, and lifelong adaptation, promising to shape how future robots learn and interact with dynamic, unstructured environments.