Study On Indoor Positioning Based On Multi-sensor Fusion

Navigation and positioning technology showed a key role in daily production and life,which function as a basic platform for many real applications.Now,Outdoor positioning technology,especially Satellite-based navigation system could provide a high-precision position result.However,there are still lots of challenge in indoor positioning domain.Although with a significant development in the area of WiFi-based,Bluetooth-based,Magnetic-based and UWB-based positioning method,it is hard to achieve a sufficient accuracy and robust.In the other hand,IMU-based positioning could obtain a high accuracy trajectory in short term positioning,but without the ability to solving the long-term position problem.In recent,researchers tend to adopt a multisensor fusion algorithm for high precision positioning,especially fusing IMU and a sensor which without accumulate error.This thesis researched two problem in the domain of multi-sensor fusion:In order to solve the positioning problem in the scene that could not set additional device,an algorithm that using magnetic measurements for loop closure detecting is adopted to reduce the accumulated error of Foot-mounted IMU based positioning method.The main contribution is proposing a method that achieves geomagnetic matching based on RANSAC method using features extracted by multi-layer FFT.Otherwise,a robust kernel function is designed for solving the low precision matching in geomagnetic matching,which achieved a higher positioning accuracy.In order to solve the Foot-mounted IMU and UWB fusion problem for complexity non-line-of-sight environment,a modified robust iterate extended Kalman filter is proposed.This algorithm dynamically adjusts the confidence to UWB measurements using the prior estimate of the current state and all the UWB measurement in currently.In experiments,this algorithm showed a higher accuracy and better robust compare to standard extended Kalman filter and robust extended Kalman filter.