Jinyu Du

Papers

Total Citations

H-Index

About

Jinyu Du is a researcher advancing the frontier of human-robot interaction and robot skill acquisition. Her work centers on enabling robots to learn complex motor skills and collaborate intuitively with humans, bridging the gap between raw sensor data and adaptive, autonomous behavior. A key contribution is her development of an adaptive multi-task human-robot interaction framework, which uses Probabilistic Movement Primitives (ProMPs) to infer and respond to human behavioral intention in real-time. This work, her most cited with 11 citations, moves beyond treating tasks independently, allowing robots to fluidly switch between collaborative actions. Du has also provided a comprehensive review of Guided Policy Search (GPS) methods, a critical class of reinforcement learning techniques that combine trajectory optimization with supervised learning to efficiently train robotic policies. Her exploration of learning in dual spaces—both task and parameter spaces—further demonstrates her commitment to robust, generalizable skill acquisition. Through these contributions, Du is shaping a future where robots are not just tools, but adaptive partners capable of learning from and with humans.