Modeling Point Uncertainty in Radar SLAM

--While visual and laser-based simultaneous localization and mapping (SLAM) techniques have gained significant attention, radar SLAM remains a robust option for challenging conditions. This paper aims to improve the performance of radar SLAM by modeling point uncertainty. The basic SLAM system is a radar-inertial odometry (RIO) system that leverages velocity-aided radar points and high-frequency inertial measurements. We first propose to model the uncertainty of radar points in polar coordinates by considering the nature of radar sensing. Then in the SLAM system, the uncertainty model is designed into the data association module and is incorporated to weight the motion estimation. Real-world experiments on public and self-collected datasets validate the effectiveness of the proposed models and approaches. The findings highlight the potential of incorporating radar point uncertainty modeling to improve the radar SLAM system in adverse environments. NOWING own pose is a fundamental problem for robotics as well as the navigation system. Recent state estimation techniques, such as simultaneous localization and mapping (SLAM), are widely used for pose estimation for navigation systems. Advancements in sensing technology have promoted the development and real-world deployment of visual and laser-based SLAM [1], [2], either independently or through sensor fusion approaches. These sensing modalities might fail well in adverse conditions, such as indoor fire scenes or outdoor snowy environments, thus blocking the application of robotics in these demanding situations.