Jyun-Ting Song

Papers

Total Citations

H-Index

About

Jyun-Ting Song is a leading researcher in computer vision and human-robot interaction, with a focus on contactless perception for safe robotic manipulation. His most notable contribution is the CORSMAL benchmark, a pioneering framework for predicting container properties—such as weight, content amount, and material—from audio-visual data. This work addresses a critical challenge in human-to-robot handovers, where opaque or transparent containers and variable materials complicate estimation. The benchmark, cited 9 times since 2022, has become a foundational resource for advancing robotic perception in dynamic, real-world scenarios. Song’s research integrates multimodal sensing and machine learning to enable robots to infer physical properties without physical contact, directly enhancing safety and efficiency in collaborative tasks. By tackling the complexities of container opacity, transparency, and material diversity, his work bridges the gap between laboratory settings and practical applications in manufacturing, healthcare, and domestic robotics. Song’s contributions are pivotal for researchers developing robust, perception-driven robotic systems, and his benchmark continues to inspire new methods for contactless property estimation in human-centered environments.