A Research Of Indoor Positioning And Navigation Using MEMS Inertial Measurement Unit And Topological Map

Indoor positioning and indoor navigation has huge potential in economy and society, due to the weakness of GNSS in indoor circumstance that there exist serious signal attenuation and multipath problems.Therefore using inertial measurement unit to do self-positioning and navigation that isn’t influenced by surrounding circumstance has became a popular research area.This thesis proposes to use MEMS IMU mounted on pedestrian’s foot to measure its motion, which can be used to estimate pedestrian’s walking trajectory. With the topological indoor map which belong to the building in which the pedestrian is walking, the position and trajectory of the pedestrian can be derived.Specificly, during the researching and constructing of a complete MEMS IMU based experimental indoor positioning and navigation system, the main work of this thesis are stated below:At first, to meet the demands for accuracy, real-time, cost in the area of indoor positioning and navigation, the needed navigation platform is designed and the completed work flow is briefly introduced.And due to the low accuracy grade of MEMS inertial measurement sensors, their error sources are devided into two categories, the deterministic biases and random noises, then an integrated calibration solution is proposed.Secondly, traditional pedestrian dead-reckoning algorithm is based on an extended Kalman filter and a pedestrian foot-stance detection method which are used to estimate pedestrian trajectory. However, this solution doesn’t take the deterministic biases into account,which leads to large divergency between the estimated trajectories and the accual ones.Therefore, after a integrated error model is established, the proposed optimal PDR solution of this thesis calibrate the deterministic biases of the sensors,the calculated deterministic coefficients and the Allan variances are used as parameters in the EKF, in that way,both kinds of errors of the sensors are compensated at the same time.According to the sensors used in the system, take the accual walking distance as the datum,the proposed solution has about 30% improvement on estimated position accuracy than the traditional method.Then, on the basis of the PDR result,a map matching algorithm based on particle filter is used to estimate the indoor position and trajectory of the pedestrian.This map matching algorithm need to use the architecture information of door,room,corridors and their spatial relationships.This thesis uses a error toleration method to extract needed architecture and to construct the topological map.Different from the existing indoor map topological solution which use the corridors as edges and turning points of the corridors as vertices.After the architectures of doors,rooms,corridors are extracted,the proposed topological solution use the pervious spaces as edges and doors as vertices.On the one hand,this map matching algorithm can use the outline of the pervious spaces as trajectory constraint which can be used in the importance sampling of the particle filter.On the other hand,the rooms are included in the pervious spaces,which conforms with the accual situation that pedestrians usually move from one room to another.Finally, the experimental map matching system is implemented on the mobile phone client, and the map matching algorithm is verified. The final experiment result showes that the accuracy of this algorithm is about ±1.5m,pedestrians can tell their indoor position with the map matching result, which proves the efficiency of this indoor positioning and navigation system.