Pedestrian Localization Algorithm Research Based On AHRS

The demand for localization and navigation is increasing with the development of modern science and technology.The other correlative techniques: GPS cannot work indoors;wireless communications need pre-install equipment.The new technique using inertial sensors is researched to meet the localization in the unfamiliar indoor environment,such as firefighting,anti-terrorism,rescue and so on.An Altitude and Heading Reference System(AHRS)developed using 9-axis sensor MPU9250 to discuss the localization algorithm in this study.An orientation algorithm based on gradient descent method is introduced in this thesis.The gradient descent method has a lower accuracy,higher speed than Kalman filters,and its parameter adjustment is more convenient.The relative displacement of pedestrian is obtained by double integrating the acceleration data.The errors of accelerometers are accumulated very quickly,so the zero-speed update algorithm is used to correct these errors.In the standing phase,the speed of the human foot is zero,so it can be inferred that the integration of acceleration data is zero.The errors of accelerometers are partially removed and the localization accuracy is improved.The height data is fused to further improve localization accuracy.Experiments show that,the acceleration measurement is much smaller than the true value when the foot touches the ground due to the limited-bandwidth.The drift after double integrating in the vertical direction is very serious.Therefore,an altimeter chip BMP180 is added in the AHRS.The double integral of accelerometer and the altimeter measurement is fused with a complementary filter.The vertical drift is removed and the localization accuracy is improved once more.Static calibration of the sensor is necessary to the localization because the offset of the gyroscope and accelerometer is not zero.The cumulative error can destroy the localization.We find that gyroscope and magnetometer calibration is particularly important,and the accelerometer calibration does not matter because of the existing of zero-speed update algorithm.The localization algorithm listed in this paper has already demonstrated for many times.The algorithm optimization and online operation will be researched deeply in the future to progress the industrial application of pedestrian localization.