Intelligent Vehicle Positioning Method Based on GPS/Lidar/Derivative Data Fusion

Haiyuan Wei,Miaohua Huang

DOI: https://doi.org/10.1109/iciscae51034.2020.9236834

2020-01-01

Abstract:Ensuring the accurate positioning of the vehicle at any time is the key to navigation and path planning in unmanned driving vehicle. Due to the changes and signal attenuation of vehicle-mounted GPS signals when they pass through electric clouds, and the reflection of GPS signals when they encounter glass curtain wall, and under underground garages, tunnels and overpasses, for example, GPS satellite signals cannot be received by vehicle-mounted GPS locator due to the obstruction of buildings. The above-mentioned situations will cause the vehicle GPS positioning signal to be unstable or even missing. And the current positioning method based on lidar has become the mainstream positioning method for unmanned vehicles. However, under bad weather conditions, the detection capability of infrared waves will be greatly attenuated, and the stability of lidar positioning will also be affected. Therefore, this paper introduces an unmanned driving vehicle positioning method based on GPS/lidar/derivative data fusion when the environment map is known. That is, when the GPS signal is good and the lidar is in good weather conditions, the vehicle can be accurately positioned by GPS/lidar/derivation fusion. While the GPS positioning signal or lidar positioning is unstable and in the event of a short absence, it is still can achieve accurate positioning of intelligent vehicles through a derivation-based approach. The experimental data and MATLAB software simulation results show that the horizontal error of vehicle positioning is within the range of 14.3cm and the longitudinal error is within the range of 8.8cm, within 5 seconds after the lidar and GPS signals are lost simultaneously.