Arjun Bhorkar

Papers

Total Citations

H-Index

About

Arjun Bhorkar is at the forefront of robotics research, pioneering methods to unify learning across diverse robotic platforms. His central focus is on **cross-embodiment learning** and **generalist navigation models**, tackling the fundamental challenge of how robots can learn from heterogeneous data sources to become more capable and adaptable. Bhorkar’s landmark work, "GNM: A General Navigation Model to Drive Any Robot" (2023, 71 citations), demonstrates that combining training data from multiple robot types can create a single, powerful navigation policy, effectively breaking the data bottleneck that limits traditional learning-based systems. He further pushes this boundary in "Pushing the Limits of Cross-Embodiment Learning for Manipulation and Navigation" (2024), investigating just how diverse a training set can be while maintaining performance. Beyond navigation, Bhorkar has made significant contributions to **offline reinforcement learning** for real-world visual navigation and developed **FastRLAP** (2023), a system enabling an RC car to autonomously practice and learn high-speed, aggressive driving from scratch using deep RL. His work is instrumental in moving toward truly generalist robots that can operate robustly in the open world.