Guosheng Hu

Papers

Total Citations

H-Index

About

Dr. Guosheng Hu is a leading researcher in autonomous vehicle perception and robotics, with a primary focus on LiDAR-based localization and 3D scene understanding. His most influential work, "STCLoc: Deep LiDAR Localization With Spatio-Temporal Constraints" (2022, 14 citations), introduces a novel deep learning framework that advances absolute pose regression by incorporating spatio-temporal constraints. This approach addresses a critical limitation of traditional map-based localization methods, which require extensive pre-built maps and struggle in dynamic environments. By directly estimating 6-DoF poses from LiDAR scans without relying on pre-existing maps, Dr. Hu's work enables more robust and efficient localization for autonomous systems. His research has significant implications for real-world deployment of self-driving vehicles and mobile robots, particularly in GPS-denied or changing environments. The STCLoc framework demonstrates how temporal coherence between sequential LiDAR frames can dramatically improve localization accuracy and reliability. Dr. Hu's contributions represent an important step toward practical, learning-based localization solutions that can operate without the infrastructure demands of conventional mapping approaches.