Changyuan Yu

Nanjing Forestry University, Tsinghua University

Papers

Total Citations

142

H-Index

About

Changyuan Yu is a researcher at the forefront of computer vision, robotics, and human-robot interaction, with a focus on bridging perception and action in complex environments. His work spans agricultural automation, human motion prediction, and reinforcement learning, demonstrating a versatile ability to tackle real-world challenges. Notably, Yu led the development of an improved F-PointNet combined with 3D point cloud clustering for mature pomegranate fruit detection and location in orchards (2022, 73 citations), advancing precision agriculture by enabling robots to autonomously identify and harvest fruit in unstructured settings. He also contributed to GIMO (Gaze-Informed Human Motion Prediction in Context, 2022, 59 citations), a seminal framework that integrates gaze data with scene context to predict human motion more accurately—critical for safe assistive robots and immersive AR/VR experiences. Most recently, Yu introduced Text2Reward (2023, 8 citations), a data-free framework that uses language models to automatically generate and shape dense reward functions for reinforcement learning, eliminating the need for specialized domain knowledge. This work promises to democratize RL by reducing development costs. With over 140 total citations and growing, Yu’s research consistently pushes boundaries, making him a rising figure in embodied AI and intelligent systems.