Xiangyu Yue

University of California, Berkeley, Chinese University of Hong Kong

Papers

Total Citations

101

H-Index

About

Xiangyu Yue is a researcher working at the intersection of machine learning, cyber-physical systems, and human motion understanding. He is best known for his foundational contributions to the Scenic probabilistic programming language, a powerful framework designed to facilitate the specification of scenarios and generation of synthetic data for machine learning-based cyber-physical systems. This work addresses critical challenges in the design pipeline, including training models to handle rare events, testing under diverse conditions, and systematic debugging — making it an invaluable tool for researchers working on autonomous systems and robotics. The Scenic language paper has accumulated over 83 citations, reflecting its significant adoption and influence within the community. More recently, Yue has expanded his research toward dynamic human motion generation, introducing SemGeoMo, a method that leverages semantic and geometric guidance to synthesize realistic human interactive motions within dynamic environments. This line of work bridges computer vision, robotics, and embodied AI, highlighting Yue's broad and forward-thinking research agenda. His contributions span both foundational programming tools and cutting-edge generative modeling, positioning him as an impactful voice in modern AI systems research.