Environmental Landmark-Based Loop Closure Detection Combining FM/WiFi/Magnetic Field Signals for Indoor SLAM

Simultaneous Localization and Mapping (SLAM) technology has been employed to automate fingerprint database construction for fingerprinting-based indoor positioning, where loop closure detection plays a critical role. Conventional methods detect loop closures by matching measurement sequences against reference templates, which suffers from low computational efficiency and limited practicality due to their dependence on overlapping paths. To reduce computational burden, some studies use specific environmental landmarks and detect loops by metrics such as inter-landmark geometric distances. However, these studies involve limited landmark types, providing insufficient constraints for cumulative positioning errors. To address these issues, this paper proposes an environmental landmark-based loop closure detection method that combines frequency modulation (FM) signal from outdoor stations with wireless fidelity (WiFi), magnetometer, and inertial sensors to enable efficient simultaneous localization and fingerprint database construction. Firstly, multiple landmarks including windows, access points (APs), magnetic anomaly points, and turning points are recognized by leveraging the significant influence of building structures/materials on radio signals and magnetic field. Secondly, the Gaussian Process Regression model is employed to expand measurement sequences into measurement matrices, thereby relaxing the requirement for overlapping walking paths through landmarks. Finally, a multi-level verification mechanism with adaptive multi-signal fusion is designed to enhance loop detection robustness. Experimental results demonstrate that the proposed algorithm can adapt to non-overlapping paths and achieve a balance between accuracy and efficiency, thus improving the positioning performance.