Simulation-Based Lane-Keeping and Traffic Avoidance System for Autonomous Vehicles

Tay, Chai Kent (2026) Simulation-Based Lane-Keeping and Traffic Avoidance System for Autonomous Vehicles. Final Year Project (Bachelor), Tunku Abdul Rahman University of Management and Technology.

Text TAY CHAI KENT_Full Text.pdf Restricted to Registered users only Download (6MB)

Abstract

The advancement of autonomous vehicles (AVs) promises a paradigm shift in transportation, offering the potential for enhanced safety and efficiency. However, the development of robust control systems capable of navigating complex and dynamic environments remains a significant challenge. This project addresses this challenge by focusing on the microscopic simulation of an autonomous vehicle’s behaviour in a working environment, with a specific emphasis on lane-keeping and traffic avoidance. A machine learning approach is employed to develop an intelligent control system that can perceive its surroundings and make real-time driving decisions. The research leverages the CARLA simulator, an open-source platform, to create a realistic and controlled testing environment. This allows for the rigorous training and validation of the proposed model without the risks and costs associated with real-world testing. The core of this work lies in the implementation and evaluation of a deep reinforcement learning (DRL) algorithm. This model is trained to interpret sensory data from the simulated environment, using camera feeds to learn optimal driving policies. The performance of the developed AV agent is systematically evaluated through a series of simulated scenarios of increasing complexity. Key performance metrics include the vehicle’s ability to maintain its lane, successfully avoid collisions with obstacles. The results of this study demonstrate the efficacy of the machine learning-based approach in achieving reliable lane-keeping and traffic avoidance capabilities. This research contributes to the growing body of knowledge in autonomous driving by providing a detailed analysis of a DRL-based control system and offering insights into the challenges and opportunities of simulation-based AV development