Zhixu Li

Fudan University, Renmin University of China

Papers

Total Citations

H-Index

About

Zhixu Li is at the forefront of embodied AI and robotic manipulation, pioneering research that bridges scene understanding and fine-grained physical interaction. His work on constructing scene-driven multimodal knowledge graphs for embodied AI (2024, 38 citations) directly addresses a critical bottleneck in robotics: enabling agents to comprehend their surroundings holistically to make intelligent, context-aware decisions. This contribution is foundational for advancing real-world robotic autonomy. Equally impactful is his innovative approach to robotic grasping, where he moves beyond conventional object-level methods to introduce 6-DoF fine-grained grasp detection grounded in part affordance (2025, 12 citations). By focusing on part-wise, shape-driven grasping, Li’s work unlocks the potential for robots to perform more dexterous, human-like manipulations—a key step toward practical assistive robotics. His research consistently targets the intersection of knowledge representation and physical action, establishing him as a leading voice in next-generation embodied intelligence. For students and researchers, Li’s work offers a compelling blueprint for how structured knowledge can empower robots to not only see, but truly understand and act upon their environment.