AbstractIndoor positioning plays a crucial role in our life and production. Typical indoor location-based applications include object or pedestrian tracking, smart logistics, intelligent manufacturing, new retail, etc. Positioning accuracy is one of the most important user requirements in any indoor positioning system. Decimetre-level or centimetre-level positioning accuracy is required for most indoor location-based applications. However, most of the current wireless indoor positioning systems cannot provide such accuracy. Even though several systems may be available to provide such accuracy, they have some other limitations or problems, such as high computation load and weak robustness. Therefore, it is necessary to develop a high-accuracy indoor positioning system which can provide real-time (without ‘perceivable’ delays) position estimations with strong robustness.
Both positioning technique/sensor and positioning algorithm should be considered for an indoor wireless positioning system for the purpose of achieving accurate position estimation. In terms of the positioning technique/sensor, Ultra-Wide Band (UWB) is characterized by high ranging accuracy, strong penetration ability, low power consumption, strong robustness to multipath effects, etc. These advantages make UWB suitable as a sensor for high-accuracy indoor positioning. In terms of the positioning algorithm, particle filter (PF) is a state-of-the-art algorithm for position estimation. PF has been widely used in the target state and position estimations because of its superiority in tackling the complicated nonlinear problems with arbitrary distributions. However, the traditional PF algorithm has some limitations and defects. When combining it with UWB for indoor positioning, it is found that they are usually failed to obtain the desired positioning results.
The overall aim of this research is to develop accurate and robust UWB and PF based indoor positioning algorithms. In this thesis, three typical problems in the current PF-based positioning algorithms are identified and analyzed, i.e., the weak robustness to the NLOS measurements in difficult indoor environments, the limitations in traditional prior position determination methods, and the particle impoverishment resulted from traditional resampling step. Moreover, the corresponding method or algorithm is proposed to tackle each identified problem.
The effectiveness of the three proposed methods/algorithms in this thesis is assessed through simulated and/or experimental tests. Besides, the positioning performance of the proposed methods/algorithms are compared with those of the state-of-the-art positioning algorithms. The test results show that each proposed method/algorithm outperforms the current positioning algorithms in solving the corresponding problem and significantly improving the positioning accuracy and robustness. Therefore, these proposed methods/algorithms are promising to be used in the applications such as people tracking in airports, object tracking in logistics, and machine guidance in Industry 4.0.
|Date of Award||Mar 2022|
|Supervisor||Ruibin Bai (Supervisor), Terry Moore (Supervisor) & Lawrence Lau (Supervisor)|
- Indoor Positioning