Papers
7
Total Citations
151
H-Index
6
About
Gijs Dubbelman is a leading researcher in robotics and computer vision, whose work centers on advancing the robustness and reliability of autonomous systems. His most significant contribution is in the field of visual SLAM (Simultaneous Localization and Mapping), where his paper "COP-SLAM: Closed-Form Online Pose-Chain Optimization for Visual SLAM" (69 citations) introduced a highly efficient, closed-form solution for pose estimation, dramatically improving the accuracy of trajectory tracking. Dubbelman has also made critical strides in addressing fundamental challenges in robot navigation, including error accumulation in visual odometry through bias compensation techniques and efficient loop closure methods. More recently, his research has expanded into machine learning for autonomous systems, pioneering methods for continual pedestrian trajectory prediction using social generative replay (20 citations) and conducting comprehensive empirical studies on domain generalization for semantic segmentation in the wild (19 citations). His work on the Eindhoven 5G Brainport Testbed (13 citations) demonstrates his commitment to real-world deployment, creating testing facilities for 5G-enabled autonomous applications. With a career spanning foundational SLAM theory to cutting-edge deep learning for robotics, Dubbelman continues to shape how robots perceive, navigate, and interact in complex, dynamic environments.
Research Focus
Key Achievements
Top Papers
- 1COP-SLAM: Closed-Form Online Pose-Chain Optimization for Visual SLAM69 citations · 2015
- 2Efficient trajectory bending with applications to loop closure20 citations · 2010
- 3Continual Pedestrian Trajectory Learning With Social Generative Replay20 citations · 2022
- 4
- 5
- 6Bias compensation in visual odometry8 citations · 2012
- 7