Haichuan Gao

Papers

Total Citations

H-Index

About

Haichuan Gao is a robotics researcher whose work focuses on bridging the gap between simulation and real-world deployment for intelligent agents. His primary research areas include robot imitation learning, reinforcement learning (RL), and sim-to-real transfer. Gao’s most impactful contribution is his work on continual robot imitation learning, where he developed the CRIL framework—a generative and prediction model that enables robots to acquire diverse skills sequentially without forgetting previously learned tasks. This work, which has garnered 17 citations, addresses a critical bottleneck in real-world robotics: the need for multi-task demonstrations to be provided all at once. He has also made notable contributions to stabilizing sim-to-real RL policies through adaptability-preserving domain decomposition, a method that resolves the inherent conflict between training stability and adaptability in domain randomization. More recently, Gao has advanced history-based RL by introducing fast counterfactual inference techniques for partially-observable tasks. His research is characterized by practical, solution-oriented approaches to fundamental challenges in robot learning, making his work highly relevant for students and researchers interested in deploying autonomous systems in dynamic, real-world environments.