Collaborative SLAM based on Wifi Fingerprint Similarity and Motion Information
This work addresses indoor positioning for users in large unknown environments, offering a pragmatic solution by leveraging existing Wi-Fi infrastructure, though it is incremental as it builds on existing SLAM and radio fingerprinting methods.
The paper tackles indoor localization by developing a collaborative SLAM system using Wi-Fi fingerprints and motion data, achieving track estimation accuracies of 0.6 meters with Tango-based PDR and 4.76 meters with step counter-based PDR in a 130m by 70m area.
Simultaneous localization and mapping (SLAM) has been extensively researched in past years particularly with regard to range-based or visual-based sensors. Instead of deploying dedicated devices that use visual features, it is more pragmatic to exploit the radio features to achieve this task, due to their ubiquitous nature and the widespread deployment of Wi-Fi wireless network. This paper presents a novel approach for collaborative simultaneous localization and radio fingerprint mapping (C-SLAM-RF) in large unknown indoor environments. The proposed system uses received signal strengths (RSS) from Wi-Fi access points (AP) in the existing infrastructure and pedestrian dead reckoning (PDR) from a smart phone, without a prior knowledge about map or distribution of AP in the environment. We claim a loop closure based on the similarity of the two radio fingerprints. To further improve the performance, we incorporate the turning motion and assign a small uncertainty value to a loop closure if a matched turning is identified. The experiment was done in an area of 130 meters by 70 meters and the results show that our proposed system is capable of estimating the tracks of four users with an accuracy of 0.6 meters with Tango-based PDR and 4.76 meters with a step counter-based PDR.