Yurou Chen

Chinese Academy of Sciences, University of Chinese Academy of Sciences, Shandong Institute of Automation

Papers

Total Citations

H-Index

About

Yurou Chen is a rising researcher at the forefront of reinforcement learning (RL) and robot imitation learning, whose work is shaping how autonomous systems master complex, long-horizon tasks. Chen’s primary contributions lie in two intersecting domains: developing adaptive algorithms for stable policy learning and enabling robots to learn intricate skills directly from human demonstration. In their most cited work, "Adaptive bias-variance trade-off in advantage estimator for actor–critic algorithms" (2023, 8 citations), Chen introduced a novel method to dynamically balance the bias-variance dilemma in policy gradients, a fundamental challenge in RL that stabilizes training and improves sample efficiency. This foundational contribution is complemented by pioneering work in hierarchical imitation learning, notably "Hierarchical Human-to-Robot Imitation Learning for Long-Horizon Tasks via Cross-Domain Skill Alignment" (2024, 6 citations), where Chen proposed a framework that extracts skill embeddings from raw human videos, enabling a general-purpose robot to solve novel, long-horizon tasks without task-specific programming. Further advancing this paradigm, "Sketch RL: Interactive Sketch Generation for Long-Horizon Tasks via Vision-Based Skill Predictor" (2023, 5 citations) introduced a method for robots to autonomously decompose complex tasks into primitive skill sequences. Chen’s research, though early in its trajectory, has already garnered significant attention for its elegant solutions to core problems in robot learning, positioning them as a key innovator in bridging human demonstration and autonomous robotic execution.