Autonomous Driving Support System with Image Processing Techniques

Abstract

In this study, an autonomous driving support system is developed using image processing and machine learning techniques. The aim of the study is to create a driving support system that can perform basic autonomous driving functions on low-cost hardware. Within the scope of the project, important driving functions, such as lane following, red light detection, sign recognition, and obstacle recognition, are addressed. While the lane following algorithm guides the vehicle by keeping it in the lane within the camera angle, the red light detection algorithm, sign recognition and obstacle recognition algorithms aim to identify the situations that need to be performed and perform the necessary activities autonomously.

Within the scope of the study, the images taken through a camera were processed using the C++ module of the OpenCV library and the specified tasks were performed using the necessary decision-making algorithms. The machine learning algorithms developed were also tested on a Raspberry Pi computer in real time and their performance was evaluated. In this way, the feasibility of autonomous driving technologies on a low-cost platform was examined, and successful results were obtained.