Radarize
| Reference number | |
| Coordinator | Örebro universitet - Örebro universitet Inst f naturvetenskap & teknik |
| Funding from Vinnova | SEK 5 300 411 |
| Project duration | April 2022 - March 2025 |
| Status | Completed |
| Venture | Advanced digitalization - Enabling technologies |
Important results from the project
The project has met its objectives. We have developed leading methods for odometry and SLAM with 2D and 4D radar, with top results on open benchmarks. For person detection, we have presented the first method for semantic segmentation in 4D radar. The results have led to several publications and international collaboration through visiting researchers.
Expected long term effects
The developed radar-based navigation and person detection methods enable robust operation in environments where cameras and lidar fail, e.g., in smoke or dust. This enhances the reliability of autonomous systems in heavy industry and rescue operations. The technology is already being integrated into a commercial product, while the results also contribute to scientific progress and strengthen Sweden’s position in robotics and AI.
Approach and implementation
The plan included software development, sensor tests under several harsh conditions, and system demonstrations. The plan was mostly kept to. The sensor tests have been useful also outside the project and for the future. The project partner responsible for close-to-hardware radar development had to leave the consortium shortly after the start, but we were able to purchase equivalent sensors and worked closely with another supplier, which meant that the result did not suffer.