Papers
16
Total Citations
365
H-Index
8
About
Archit Sharma is a robotics and machine learning researcher whose work sits at the intersection of robotic learning, reinforcement learning, and large-scale data-driven methods. His research focuses on enabling robots to acquire generalizable, real-world manipulation skills through a combination of imitation learning, reinforcement learning, and foundation model approaches. Sharma has made significant contributions to large-scale robotic learning infrastructure, most notably through his involvement in the Open X-Embodiment project (119 citations) and the DROID dataset (108 citations), both of which advance the development of general-purpose robotic models trained across diverse environments and hardware platforms. His work on SERL (31 citations) provides practical software tools for sample-efficient real-world robotic reinforcement learning, lowering the barrier for researchers entering the field. He has also explored language-guided robot correction through "Yell At Your Robot" (34 citations), enabling intuitive on-the-fly human intervention during deployment. Earlier work on unsupervised off-policy reinforcement learning demonstrated how robots could develop skills without hand-crafted reward functions, reflecting a consistent theme in his research: reducing human engineering effort while expanding robotic capability. Collectively, Sharma's contributions are shaping how the field scales robotic learning toward practical, generalizable autonomy.
Research Focus
Key Achievements
Top Papers
- 1
- 2DROID: A Large-Scale In-The-Wild Robot Manipulation Dataset108 citations · 2024
- 3Yell At Your Robot: Improving On-the-Fly from Language Corrections34 citations · 2024
- 4SERL: A Software Suite for Sample-Efficient Robotic Reinforcement Learning31 citations · 2024
- 5
- 6
- 7
- 8Waypoint-Based Imitation Learning for Robotic Manipulation8 citations · 2023
- 9
- 10