TraMap: SLAM-Based Trajectory Generation and Optimization for Emergency Scenarios

Under the background of frequent accidents in recent years, how to determine the location of rescuers has attracted extensive attention. Wi-Fi and inertial sensors are common sources for indoor positioning and trajectory generation. However, their popularity is limited by the error drift within Inertial Navigation System (INS), fluctuations in Wi-Fi Received Signal Strength Indicators (RSSI), and the need to deploy nodes in advance. In this paper, we design an indoor localization and trajectory generation system, “TraMap", which generates trajectories from Pedestrian Dead Reckoning (PDR) and corrects them by Fine Time Measurement (FTM). We present a trajectory optimization method to optimize the PDR-generated trajectories cyclically. We also propose a real-time Unscented Particle Filter (RUPF) to fuse the FTM ranging results and trajectories of the PDR. We evaluate the performance of TraMap in actual scenarios. The experimental results show that TraMap improves the positioning error by more than 35% compared with the traditional Wi-Fi fingerprint-based localization.