Advancing indoor multi-person localisation system based on sensor fusion method

Abstract

Indoor positioning systems (IPS) have garnered increasing attention in the field of positioning research in emergency services such as firefighting scenarios. The capability to deliver precise and comprehensible positioning information for multiple firefighters in harsh environments is a promising technology. It will effectively save their lives via timely and accurate location information for evacuation and reinforcement. The sensor fusion-based dead reckoning (DR) method is one of the typical techniques in IPS. Due to its little reliance on layout knowledge and pre-installed positioning hardware in the building, it is regarded as one of the most promising methods for IPS in firefighting. Existing research on DR has not adequately addressed the challenges of positioning accuracy, surrounding reconstruction, and multi-person positioning in firefighting scenarios. In order to address these problems, this thesis explores advanced DR based multi-person localisation and mapping. The research work consists of five associated studies that aim to answer the formulated research questions. The first three studies explore the novel approaches in gait analysis-based heading estimation, dual foot synergistic step detection and dynamic minimum stride length constraint-based positioning optimisation. The objective of these studies is to improve the precision of positioning by optimising parameters in the DR calculation process. The next study presents a geometry algorithm that utilises a polar projection strategy to determine the coordinates of map points and reconstruct the user's surrounding map. The last study explores an innovative approach for integrating multiple trajectories via online magnetic fingerprint matching. By doing so, the position of each individual is updated by combining fingerprint information. This thesis conducts experiments to evaluate the performance of the systems proposed in each study. Each experiment is tailored with specifically designed realistic indoor scenarios, data collection hardware, and evaluation metrics. The quantitative assessment results illustrate improved positioning accuracy in comparison to conventional methods. The displayed trajectory and map demonstrate accurate results that exhibit high consistency with the ground truth.

Date of Award	Jul 2024
Original language	English
Awarding Institution	University of Nottingham
Supervisor	Boon Giin Lee (Supervisor), Matthew Pike (Supervisor) & Liang Huang (Supervisor)