Mingfei Sun

Papers

Total Citations

H-Index

About

Mingfei Sun is a rising researcher at the forefront of human-robot interaction and imitation learning. Their work centers on two pivotal challenges: enabling robots to faithfully replicate complex human behavior and efficiently capturing human preferences for robot task alignment. Sun’s most influential contribution, “Imitating Human Behaviour with Diffusion Models” (2023, 23 citations), pioneers the use of diffusion models—typically celebrated in text-to-image generation—as observation-to-action models for sequential environments. This approach uniquely captures the stochastic, multimodal nature of human behavior, addressing a critical gap in traditional imitation learning. In parallel, Sun’s “FARPLS” system (2024) introduces a feature-augmented trajectory preference labeling tool, designed to reduce cognitive load on human labelers while improving the quality of preference elicitation for robot training. Though early in their career, Sun’s work bridges generative AI and robotics, offering scalable solutions for learning from human demonstration and feedback. Their research holds significant promise for developing more intuitive, human-aligned autonomous systems, marking Sun as a notable innovator in the field.