The Research Of Multi-sensor Pedestrian Dead Reckoning Method And UKF Fusion Algorithm

With the rapid rise of mobile Internet,positioning and navigation technology is used in many fields.In the outdoor environment,Global Positioning System(GPS)can obtain the location of user with satellite signal well,but in the indoor environment,the satellite signal is obstructed and it's hard to acquire accurate location.At present,Inertial Navigation Systems(INS)relied on the Inertial Measurement Unit(IMU)has become the main autonomous navigation system,but the high-precision IMU is bulky and expensive that is difficult to use widely.In recent years,the smart mobile devices have been used popularly in people's lives,and a majority of devices have been embedded by IMU and other sensors.Therefore,this paper proposes a multi-sensor indoor pedestrian dead reckoning method based on the low-cost sensor of smart mobile device.Aiming at solving these problems of low cost sensor,the corresponding model of error correction is designed which is divided into three sections as followed:1?The initial alignment: The initial alignment can make the navigation coordinate system coincide with the coordinate system which is described by INS and give a correct initial value when the computer is working normally.It is known that IMU is easily disturbed by other components based on smart mobile device,and therefore this paper discuss the introduction of unscented Kalman filter in the fine phase of initial alignment to get the accurate initial information,and also fuses the multi-sensor data to correct error.2?Movement state detection model: It's important to collect the correct state information with IMU for using pedestrian dead reckoning method to solve highprecision position,speed and attitude information when the pedestrian at the sport state.When the pace state of pedestrian is quite different,it is difficult to obtain the correct stance information of pedestrian only by accelerometer.In this paper,the motion data obtained by accelerometers and gyroscopes is used to set four threshold conditions to detect the state of pace based on multi-sensor hardware platform of intelligent mobile devices.3?Multi-sensor pedestrian dead reckoning method: Based on the initial alignment and movement state detection model,aiming at to get rid of noises interfused and to solve the problem of error accumulation in the multi-sensor working and solution,this paper proposes a zero velocity update,zero angular rate update and magnetometer fusion method based on unscented Kalman filter to correct the heading angle and speed error effectively.After several experiments and data analyses,the proposed method of this paper can get the average position error of about 1.57% of the Omni distance and it can be better fulfill the requirements of indoor positioning.