Research On The Indoor Inertial Navigation Method Based On Vision-aided

Indoor location service has become an urgent demand recently,the indoor positioning technology has been developed rapidly,and it played a great role in the indoor environment such as cinema,shopping mall,hospital,exhibition hall,our lives are impacted by indoor location.The main positioning methods are GPS(Global Positioning System),inertial navigation system,visual.GPS accuracy is high,errors do not accumulate with time,but GPS signals are easily blocked by buildings,therefore,GPS cannot be used in indoor environment.The inertial navigation system uses gyroscopes and accelerometers to measure the acceleration and angle rate,in order to obtain the position and velocity.INS(Inertial Navigation System)is an integral inference navigation system,so the error will increase with time.INS measuring devices has long shift,so it is difficult to work alone.The improvement of computer operation ability and digital image processing algorithm have became more and more mature,in addition,the cost of industrial camera is cheaper than before,therefor,visual positioning method is used to extract the information of the carrier's running state by machine vision widely.With the maturity of various information fusion technologies,integrated navigation technology with complementary characteristics is an effective way to improve the performance of INS.The working principle and error model of inertial navigation system are studied deeply.To decrease the error of INS,a visual/INS integrated navigation which is based on navigation database is proposed.Firstly,a navigation database based on SQL software is proposed.The visual algorithm of camera calibration is proposed to estimate the parameters of camera.In addition,in order to achieve visual template matching correctly,a gray value matching algorithm and SURF feature point detection and tracking algorithm are studied.A fusion algorithm based on Kalman is studied too.The error of INS is taken as the filtering states,and the optimal estimation is realized through the filtering equation.To solve the problem of delay,a delay calibration method for visual and inertial sensors is proposed.Models are simulated based on MATLAB and the visual software HALCON,the simulation results are analyzed from two aspects: position error and speed error.They show that the integrated positioning error based on visual aided is smaller than inertial navigation working alone.