Lars Hammarstrand

Lars Hammarstrand is an Associate Professor at Chalmers University of Technology specializing in the Signal Processing research group. His work integrates model-based Bayesian statistics with deep machine learning for applications in visual localization, sensor fusion, and autonomous systems, with emphasis on robustness in real-world environments.

His research focuses on bridging Bayesian inference and deep learning to solve challenges in autonomous vehicle perception. Key areas include visual localization under appearance changes, radar-camera sensor fusion, and out-of-distribution detection for safety-critical systems. Recent work explores neural radiance fields for radar, semi-supervised learning for mapping, and probabilistic hierarchical classification to address real-world uncertainties in autonomous driving.

Analysis of his 2020-2025 publications reveals a trajectory toward unifying geometric and semantic understanding in autonomous systems. His work demonstrates increasing integration of neural radiance fields with traditional filtering techniques, while advancing open-set recognition capabilities. Notable contributions include road geometry estimation frameworks, extended object tracking with PHD filters, and methods to mitigate data leakage in localization benchmarks.

No scientific awards were mentioned in the provided materials.

Hammarstrand has contributed to academic supervision methodology through his publication on improving master's thesis supervision efficiency, though specific student names are not listed. The provided information contains no details about research grants or funding sources.

He operates within Chalmers University's Signal Processing research group, which develops advanced algorithms for automotive perception systems, focusing on sensor fusion techniques that combine radar, camera, and motion data for robust environmental understanding in autonomous vehicles.