Wi-Loop SLAM: Loop Closures with Wireless Sensing in Multipath SLAM

Loop closure detection is an important aspect of simultaneous localization and mapping (SLAM) to correct long-term drift by finding overlapping trajectories. However, in multipath SLAM using radio signals, due to the limitation of low-resolution sensors and few features, most of the existing works have not fully explored loop closure. In this paper, we propose a loop closure detection algorithm for multipath SLAM, Wi-Loop SLAM, that can reduce cumulative errors. To recognize previously visited places, we use the multipath parameters extracted from received signals as matching features without introducing new information. We use a delayed selection strategy to ensure that the most appropriate pairing is selected, and a confirmation of location drift is required to minimize unnecessary computation. Finally, we derive a Bayesian model to perform loop optimization with the particle-based sum-product algorithm (SPA). Simulation results show that the proposed Wi-Loop SLAM can correct the state estimate drift in the environment where propagation paths are often obscured.