Ram Ramrakhya

Papers

Total Citations

H-Index

About

Ram Ramrakhya is a researcher specializing in embodied AI and visual navigation, with a particular focus on enabling autonomous agents to intelligently navigate complex, unfamiliar environments. His most notable contribution, **PIRLNav** (2023), addresses the challenging problem of ObjectGoal Navigation — training virtual robots to locate specific objects within previously unseen environments. Recognizing the limitations of pure imitation learning approaches, Ramrakhya and his collaborators developed a hybrid training paradigm that combines behavior cloning from human demonstration datasets with reinforcement learning fine-tuning, achieving significant performance improvements over prior methods. This work has garnered over 56 citations, reflecting its meaningful impact on the embodied AI research community and its relevance to robotics, computer vision, and autonomous systems. Ramrakhya's research sits at the intersection of human-guided learning and autonomous policy optimization, advancing how AI agents can leverage human expertise as a foundation while continuing to improve through self-directed exploration. His contributions are particularly valuable for researchers working on household robotics, simulation-to-real transfer, and scalable training strategies for navigation agents, making him a noteworthy emerging voice in the embodied intelligence landscape.