A Correlation-Based Method to Use Street Profile Data for Accurate Localization of Vehicles

To enable the concept of a cloud-aided proactive suspension control, a precise localization of the vehicle on the street is needed and the accuracy must be significantly better than what global navigation satellite systems can provide. In this paper, we present a method to localize the vehicle in longitudinal direction using the information of the vertical street profile. The method is based on the correlation of a reference street profile from the cloud server with the observed profile gained through the vehicle's sensor data to determine the position of the vehicle relative to the cloud profile. The suitability of the algorithm is demonstrated on a quarter vehicle test bench using real sensor data and street profiles recorded on real roads. Given an accurate reference profile, this setup achieves a very precise localization on a variety of different road types with a mean absolute position error of about 1 cm and the corresponding mean absolute time error of around 1ms. We also tested the method for robustness and found that local disturbances in the street profile or noisy observer data did not impair the performance. However, the algorithm is rather sensitive to inaccuracies in the measured vehicle speed. Even though we developed the method with respect to the intended application in proactive suspension control, it might also be of interest in other areas like autonomous driving, where high-precision localization is required.     «

To enable the concept of a cloud-aided proactive suspension control, a precise localization of the vehicle on the street is needed and the accuracy must be significantly better than what global navigation satellite systems can provide. In this paper, we present a method to localize the vehicle in longitudinal direction using the information of the vertical street profile. The method is based on the correlation of a reference street profile from the cloud server with the observed profile gained t...     »