Radar Localization and SLAM
Motivation
Localization of mobile robots in extreme situations is an important future topic. Radar shows its strengths in situations, where lasers and other conventional sensors fail. This enables us to determine the position of the robot even under the worst visibility conditions like dust, fog and smoke.
Goal
The IGMR focuses on two approaches. First Radar SLAM, pure navigation and mapping with radar, second, localization with radar in known environments. The goal of both variants is to achieve a real-time capable and robust position determination of the robot.
Approach
For the radar SLAM we use probabilistic methods based on Probabilistic Iterative Correspondence, short pIC. Here the highly noisy input signal is added to the corresponding sections of the map via a probability distribution. We rely in particular on efficient prefiltering.
For radar localization, the robot orients itself on a laser-generated map to increase the overall accuracy. We rely on laser-native particle swarming methods and the methodical approximation of the radar signal to typical LiDAR scans.
Sectors
- Situations wtih bad visibility conditions
- Mining industry
- automotive industry
- Rescue services: like fire brigade
- exploration of complex environments for example Forests, demolition site
Funding