Driving behavior prediction of a surrounding moving vehicle using a map for autonomous driving

Abstract

In recent years, autonomous driving for the purpose of reducing traffic accidents is studied. The recognition of surrounding environment is one of the important technologies to realize it. In our work, position, velocity and accretion of surrounding moving vehicle is recognized by LIDAR and Millimeter Wave Radar. Then, the predicted trajectory of the moving vehicle is generated by using the recognized information. When there are multiple roads such as intersection, there is a problem that it cannot be predicted which road moving vehicles will go a few seconds later. This work proposes a method to predict which road the moving vehicles can go next by predicting lateral acceleration calculated the curvature of the road from the map and recognition information. Simulation experiment is performed a straight line and a left turn on a road, as a result it is possible to predict in advance.