Dedong Liu

Papers

Total Citations

H-Index

About

Dedong Liu is a researcher advancing the field of robotic perception, with a primary focus on **3D multi-object tracking (MOT)** for autonomous systems. His work addresses a critical challenge in mobile robotics: enabling platforms to perceive and predict the motion of surrounding objects in real-time for safe navigation and planning. Liu’s major contribution is the development of the **Polyhedral framework**, introduced in his highly cited paper “Poly-MOT” (52 citations). This framework innovatively replaces traditional single-metric data association with a polyhedral approach, integrating multiple similarity measures and physical models to significantly improve tracking robustness in complex environments. Building on this, his subsequent work **Fast-Poly** (19 citations) tackles the dual challenges of accuracy and latency consistency, proposing a filter-based method that achieves high-speed, reliable tracking suitable for real-world deployment. Liu’s research has direct implications for autonomous driving, warehouse logistics, and service robots, where understanding dynamic surroundings is paramount. By addressing both theoretical limitations and practical performance bottlenecks, Dedong Liu is helping to bridge the gap between cutting-edge 3D perception algorithms and their real-time application in embodied AI systems.