Shao-Hua Sun

Papers

Total Citations

H-Index

About

Shao-Hua Sun is a rising force in robot learning, whose work bridges the critical gap between sample efficiency and real-world deployment. His research centers on meta-reinforcement learning and imitation learning, with a particular focus on enabling robots to acquire complex, long-horizon behaviors from limited data. In his highly cited work on "Skill-based Meta-Reinforcement Learning" (2022, 11 citations), Sun tackles the fundamental challenge of sample inefficiency in deep reinforcement learning, proposing a framework that allows agents to leverage reusable skills for rapid adaptation—a key step toward making robot learning feasible on physical systems. More recently, his 2023 paper on "Diffusion Model-Augmented Behavioral Cloning" (3 citations) pushes the boundaries of imitation learning by using generative diffusion models to better capture expert distributions without requiring environment interaction. This approach addresses a core limitation in observational learning: modeling complex, multimodal behaviors from demonstrations. Sun’s contributions are particularly notable for their practical orientation—he focuses on algorithms that can actually run on real robots, not just in simulation. As the field of robot learning races toward general-purpose embodied intelligence, Sun’s work on sample-efficient, skill-based, and demonstration-driven methods positions him as a key architect of the next generation of capable, data-efficient robotic systems.