| Indoor positioning technology has been widely used in shopping mall navigation,smart home,personnel search and rescue and other fields.It has immeasurable commercial value and very broad application prospects.Global Navigation Satellite System(GNSS)can provide good positioning accuracy for outdoor positioning,but the invisible satellite signal in an indoor environment makes it difficult to obtain indoor location information,which makes it difficult to meet the needs of indoor positioning.Therefore,a hot research topic came out at home and abroad,which is how to realize the positioning service to satisfies the indoor environment demand.Although many applications for indoor positioning technology have been developed at this stage,a single indoor positioning technology works ineffective due to its own limitations in the complex and changeable space-time indoor environments.Refer to the outdoor integrated navigation,fusion positioning technology has become an effective method to solve this problem.In this paper,the study will be conducted from the two positioning technologies respectively,which are WiFi fingerprint positioning and pedestrian track estimation(PDR),the existing problems in these two aspects will be reasonably proved.Also,the method of information mixed-up from two positioning technologies will be studied.The research content of the article are as follows:Firstly,aim at the time-wasting and laborious problem of WiFi fingerprint when collect the strength of off-line signal data,an improved kriging interpolation algorithm is proposed to quickly build an offline fingerprint database.At first,the particle swarm optimization algorithm is used to fit the variogram model more accurately,and then the data of known points and the fitted variogram model are interpolated at the unknown points to obtain the offline fingerprint database.The experimental results show that the improved kriging interpolation algorithm has better interpolation effect,and its database building effect is close to the traditional grid acquisition method.Secondly,for the WiFi fingerprint online positioning and matching algorithm,an online positioning algorithm for improving the sorting of reference points is proposed to reduce the influence of the dynamic change of WiFi signal strength and physical location information.Experimental results show that the cumulative distribution probability of the proposed algorithm's positioning error within 3m is 81%,which has higher positioning accuracy than other traditional machine learning positioning matching algorithms.Thirdly,for the data processing problem of the smartphones' built-in sensor,a gait detection algorithm based on the finite state machine and the course angle smoothing using the Kalman filter are proposed to perform PDR calculation;extracting the feature of acceleration data through sliding window,and the XGBoost classification algorithm is used for model training to identify the pedestrian motion state.Experimental results show that the proposed gait detection algorithm has strong anti-interference ability and small heading estimation error;the recognition accuracy of the XGBoost classification algorithm is 95.3%,which can effectively use the recognition results to assist PDR positioning.Finally,to solve the WiFi fingerprint positioning's result jumping problem and error accumulation of PDR positioning,an optimized WiFi/PDR fusion positioning algorithm is proposed.The step length information in the PDR solution is used to constrain the WiFi fingerprint positioning,and then the extended Kalman filter is used to fuse the constrained WiFi location result and the PDR solution result.Taking experiments on normal walking and fast walking on WiFi positioning,PDR positioning,and fusion positioning,which carries out that the optimized fusion positioning result is closer to the actual trajectory.The fusion positioning accuracy under normal walking and fast walking is 0.86 m and 0.98 m respectively,the accuracy and stability are significantly better than the results of WiFi fingerprint positioning,PDR positioning and EKF fusion positioning. |
PDF Full Text Request