Papers
6
Total Citations
122
H-Index
6
About
Jesse Zhang is a leading researcher at the intersection of reinforcement learning, robotics, and safety-critical AI systems. His work focuses on enabling robots to learn efficiently from limited data while operating reliably in real-world environments. Zhang’s most impactful contribution is COG (2020, 38 citations), which pioneered connecting new skills to past experience through offline reinforcement learning, dramatically reducing the need for task-specific data collection. He also developed Cautious Adaptation (2020, 26 citations), a framework for safe RL deployment in hazardous settings like autonomous driving, addressing a critical barrier to real-world adoption. Zhang co-created REPLAB (2019, 24 and 21 citations), a reproducible, low-cost robotic arm benchmark that has become a standard platform for vision-based manipulation research. His more recent work includes RoboCLIP (2023), demonstrating that robot policies can be learned from just a single demonstration, and SPRINT (2024), which scales policy pre-training by automatically relabeling language instructions. Through these innovations, Zhang has accumulated over 120 citations, establishing himself as a key figure in making robotic learning more sample-efficient, safe, and accessible.
Research Focus
Key Achievements
Top Papers
- 1
- 2Cautious Adaptation For Reinforcement Learning in Safety-Critical Settings26 citations · 2020
- 3REPLAB: A Reproducible Low-Cost Arm Benchmark for Robotic Learning24 citations · 2019
- 4REPLAB: A Reproducible Low-Cost Arm Benchmark Platform for Robotic Learning21 citations · 2019
- 5RoboCLIP: One Demonstration is Enough to Learn Robot Policies7 citations · 2023
- 6SPRINT: Scalable Policy Pre-Training via Language Instruction Relabeling6 citations · 2024