Gan Sun

Chinese Academy of Sciences, Shenyang Institute of Automation

Papers

Total Citations

H-Index

About

Gan Sun is a leading researcher in lifelong and open-ended learning for robotic perception, with a focus on integrating visual and tactile modalities. His work addresses a critical challenge in robotics: enabling systems to continuously learn and adapt from new sensory data without forgetting previously acquired knowledge. Sun’s most influential contribution is his pioneering framework for lifelong robotic visual-tactile perception learning (2021, 34 citations), which established a foundation for how robots can fuse and interpret touch and sight over extended interactions. He advanced this with a novel lifelong visual-tactile spectral clustering method (2022, 18 citations) that improves object recognition by dynamically grouping sensory inputs. Sun has also pushed the boundaries of autonomous visual systems with his work on open-ended online learning (2023, 7 citations), where robots collect and perceive visual data in real time, mimicking human-like learning. Most recently, he introduced I3DOD (2023), an incremental 3D object detection framework that uses prompting to overcome catastrophic forgetting in class-incremental scenarios—a key hurdle for deploying detection models in autonomous driving and augmented reality. Through these contributions, Sun is shaping the future of adaptive, lifelong robotic intelligence.