Breakfast seminar: Beyond the Ground Plane: 3D Radar Localization from Range and Bearing with Magnus Oskarsson

Breakfast is served at 09:00, the presentation will start shortly after. For breakfast, please sign up no later than May 22^nd by using this link.

Radar is widely used for robust localization and mapping, but many existing methods still rely on planar motion assumptions and produce essentially 2D maps and poses. This talk explores what becomes possible when radar range and bearing measurements are treated as genuinely 3D geometric constraints. I will present how we can model such estimation problems, and show two examples, triangulating 3D target points from multiple posed radar measurements, and estimating the full 6-DOF pose of a radar sensor from range and bearing observations. In both cases, the key idea is to exploit the underlying geometry of the measurements to obtain efficient and globally optimal or least-squares optimal solutions. The resulting formulations connect radar perception with tools from geometric estimation, eigenvalue problems, and maximum likelihood estimation.

Magnus Oskarsson is Senior Lecturer in Computer vision and Machine Learning at the Center for Mathematical Sciences. He works on methods in computer vision and machine learning, applied to navigation, localization and 3D reconstruction.