交差点での右左折における目標操舵角生成手法

抄録

Autonomous driving in highway uses lane markers detection with stereo camera, LiDAR, and radar. However, when we drive in the city, an intersection has no lane markers, so we cannot detect them. Therefore, we need control method without using lane markers detection. We propose a method to drive right/left cornering in an intersection with smoother steering. We use self-position data from GPS and map information composed of nodes and links of road. First, we generate target steering angle, which follows vehicle trajectory to clothoid curve from map information. Next, we calculate gain value, which leads commanded steering angle to target steering angle by least square method. Simulation results showed that we can generate commanded steering angle, which makes vehicle trajectory to clothoid curve in any corner with gain value we calculated.