Jinming Li

University of Shanghai for Science and Technology

Papers

Total Citations

H-Index

About

Jinming Li is at the forefront of embodied artificial intelligence, pioneering the integration of foundation models with robotics to create machines that can perceive, reason, and act in the physical world. His key research areas span visuomotor control, robotic manipulation, and the development of scalable, data-efficient Vision-Language-Action (VLA) models. Li’s major contributions include a comprehensive survey that charts the path from robotics with foundation models toward true embodied AI, establishing a critical framework for the field. He has also advanced the scalability of robotic learning, demonstrating that Diffusion Policy can be effectively scaled to a Transformer with 1 billion parameters, a breakthrough that pushes the boundaries of model size and performance in manipulation tasks. Recognizing the practical limitations of large models, Li introduced TinyVLA, a fast and data-efficient VLA architecture that addresses the critical challenges of inference speed and extensive pre-training. Though his most-cited works are recent (2024-2025), they have already garnered significant attention, with each paper accumulating 3-4 citations in a short span, signaling a rapidly growing impact. Li’s work is shaping the next generation of intelligent, general-purpose robotic agents.