Research On Indoor Combined Positioning Technology Based On Image And Inertial Navigation Data

Indoor location services are playing an increasingly important role in production and life,and various indoor positioning technologies have also developed rapidly.Among these indoor positioning technologies,visual positioning technology and inertial navigation positioning technology have their own advantages,which have become important directions in indoor positioning research.However,the system real-time performance of the visual positioning technology and the robustness when it is affected by non-line-of-sight factors need to be improved.The problem of reduced positioning accuracy caused by the accumulation of errors in inertial navigation positioning technology also needs to be solved urgently.In order to overcome the shortcomings of single positioning technology and realize the complementary advantages of different positioning technologies,this paper studies the indoor combined positioning technology based on image and inertial navigation data.The main research works are as follows:(1)This paper analyzes the causes of poor real-time performance and the causes of positioning errors of visual positioning technology based on image matching,and proposes an improved image matching positioning method.Firstly,an accelerated robust feature algorithm is applied to image feature extraction to obtain 64-dimensional feature descriptors.Then the inertial navigation positioning data is used as auxiliary information to establish a landmark library search list that can narrow the landmark library search range,and the fast nearest neighbor search algorithm is used to accelerate feature matching to obtain the landmark image with the highest matching degree.Finally,the matching points information in the search list is fully used to perform visual weighted centroid calculation to obtain the final visual positioning output.Experimental results show that the speed of feature extraction,feature matching and retrieval of the improved method has been increased by 14%,52%,60%,and the positioning error has decreased by 36%.(2)Aiming at the problem that image matching positioning technology cannot provide positioning output when it is influenced by non-line-of-sight and the accumulation of inertial navigation positioning errors,a combined positioning method of image and inertial navigation data fusion is proposed.Firstly,based on the characteristics of image and inertial navigation positioning data,a Kalman filter is selected as the fusion algorithm of image and inertial navigation positioning data.Then,a tightly combined data fusion method is designed and a corresponding system equation is constructed to perform target positioning and inertial navigation error feedback correction.The experimental results prove that the combined positioning technology can only rely on the inertial navigation module for positioning under non-line-of-sight conditions,and can use the tight combination feedback correction mechanism to suppress the inertial navigation error accumulation under the line-of-sight conditions.The system robustness and the positioning accuracy of the combined positioning technology are better than the single positioning technology.(3)A combined positioning system that can be used for autonomous traveling equipment positioning is designed by using positioning method based on the fusion of image and inertial navigation data.After experimental tests,we can find that the host computer application in the combined positioning system can get target position with or without visual positioning output.