Stop Sign and Stop Line Detection and Distance Calculation for Autonomous Vehicle Control

Abstract

Environmental perception plays a crucial role in autonomous driving vehicle speed control. Autonomous vehicles must follow the traffic rules indicated in traffic sings. In this paper, a novel method is proposed for detection of stop sign and calculating the distance, which is an essential parameter in controlling the longitudinal velocity of an autonomous vehicle. As the vehicle moves closer to the stop sign, the stop sign falls out of the field of view of the camera, making it tough to bring the vehicle to stop at the desired distance from the sign. Hence, information on the position of stop line is essential to know where exactly to stop the vehicle. Stop sign detection is carried out using AdaBoost cascade classification based on three different feature types- Haar-like features, Local Binary Patterns (LBP) and Histogram of Oriented Gradients (HOG). The performance results of all the three classifiers are analyzed and compared to determine which one performs the best. To find the stop line a classic computer vision algorithm is proposed. The distance to stop sign and stop line is estimated in real time so that a decelerating torque can be applied accordingly to slow down the vehicle and eventually bring it to a complete standstill.