Rachel Fong

Papers

Total Citations

3,294

H-Index

About

Rachel Fong is a leading researcher in robotics and reinforcement learning, best known for her pioneering work in bridging the simulation-to-reality gap. Her key research areas include domain randomization, transfer learning, and sample-efficient reinforcement learning. Fong’s most impactful contribution is the introduction of domain randomization, a simple yet powerful technique that trains neural networks on randomized simulated images to achieve robust transfer to real-world environments. Her seminal 2017 paper on this topic has garnered over 2,700 citations, fundamentally reshaping how roboticists approach sim-to-real transfer. In parallel, she co-developed Hindsight Experience Replay (HER), a breakthrough method that enables sample-efficient learning from sparse, binary rewards—eliminating the need for complex reward engineering. This work, with over 350 citations, has become a cornerstone for tackling challenging reinforcement learning problems. Fong’s innovations have accelerated robotic research by making simulated training data more practical and accessible, directly impacting fields from autonomous manipulation to industrial automation. Her contributions continue to inspire new generations of researchers seeking to deploy learned policies on physical hardware.