Arjun Bhardwaj

Papers

Total Citations

H-Index

About

Arjun Bhardwaj is a pioneering researcher at the intersection of socially intelligent robotics and data-efficient machine learning. His work bridges two critical domains: creating AI agents that can perceive and respond to human social cues, and developing algorithms that learn rapidly from minimal data. Bhardwaj’s most influential contribution is the SARA (Socially-Aware Robot Assistant) system, introduced in his 2016 paper (80 citations), which represents one of the first embodied intelligent personal assistants capable of analyzing a user’s visual behavior (head and face movements), vocal acoustics, and conversational strategies to estimate rapport. This system could dynamically adjust its own visual and verbal responses to build better human-robot relationships. More recently, Bhardwaj has advanced the frontier of meta-reinforcement learning with his PACOH-RL algorithm (2024), which meta-learns probabilistic priors for dynamics models, enabling robots to adapt control policies to entirely new environments with remarkably few interaction trials. This work addresses the critical challenge of data efficiency in robotics, where collecting real-world experience is expensive. Bhardwaj’s research uniquely combines social awareness with sample-efficient learning, positioning him as a key figure in developing robots that are both socially adept and practically deployable.