RaCo-SLAM: A Physics-Informed 4D Radar SLAM with Co-Visibility Consistency Factor
Zishun Deng, Wanbiao Lin, Can Li, Teng Wang, Chao Guo, Jiawei Shen, Lei Sun
AI summary
Problem
High-precision SLAM with 4D mmWave radar is hindered by extremely sparse and noisy point clouds, making robust feature extraction and long-term global consistency difficult without relying on fragile explicit loop-closure methods.
Approach
The system adaptively extracts stable static features using a physics-informed probabilistic model that fuses Doppler, RCS, and intensity data, while a novel co-visibility consistency factor directly minimizes point-to-point registration errors between overlapping keyframes to correct long-term drift without explicit loop closures.
Key results
- Complete real-time 4D radar SLAM framework with physics-informed adaptive feature extraction
- Novel co-visibility consistency factor for continuous global drift correction on standard CPUs
- State-of-the-art accuracy and robustness across diverse challenging real-world datasets
- Real-time performance exceeding 40 Hz on a standard CPU without GPU acceleration
Why it matters
It enables robust, all-weather autonomous navigation for ground robots and vehicles in adverse conditions where optical sensors fail, using only affordable 4D radar hardware on standard CPUs.
Abstract
Robust all-weather localization is a critical ca- pability for autonomous systems. While 4D mmWave radar offers superior resilience to adverse environmental conditions compared to LiDAR and cameras, its application in high- precision Simultaneous Localization and Mapping (SLAM) is hindered by significant challenges, including severe point cloud sparsity, complex noise characteristics, and the prevalence of dynamic objects. To address these issues, we propose RaCo- SLAM, a robust and real-time 4D mmWave radar SLAM framework with co-visibility consistency. This framework fea- tures a novel physics-informed probabilistic model for adaptive feature extraction from sparse and noisy point clouds. For global consistency, we introduce a co-visibility consistency factor (CoVC factor) into the global optimization, moving beyond conventional loop-closure methods. This factor directly minimizes point-to-point registration errors to enforce global consistency and is designed for parallel real-time execution on a standard CPU. Comprehensive evaluation on diverse and challenging real-world datasets demonstrates state-of-the- art accuracy and robustness, achieving real-time performance exceeding 40 Hz on a standard CPU. To benefit the com- munity, the code and collected dataset will be released at https://github.com/sudo-robot0/RaCo-SLAM.