Multi-Sensor Fusion and V2V Collaboration in Autonomous Vehicles: A Comprehensive Framework for Self-Driving

Self-driving cars, also known as autonomous vehicles, represent a significant advancement in transportation technology. These vehicles utilize a combination of LiDAR sensors, cameras, and artificial intelligence to navigate without human intervention. The development of self-driving technology promises to enhance road safety, reduce traffic congestion, and decrease vehicle collisions. As the demand for autonomous vehicles increases, so does the need for robust systems that can reliably interpret complex environments, necessitated by the current challenges of navigating through dynamic, uncertain, and safety-critical scenarios in real-world driving. Key components of autonomous driving include lane detection, traffic sign recognition, and vehicle tracking, all of which are crucial for ensuring safe and efficient navigation. In addition, vehicle-to-vehicle collaboration plays an essential role in facilitating communication among autonomous vehicles, enabling them to make informed decisions based on real-time data from their surroundings. This paper presents a multi-modal autonomous driving framework designed for robotic vehicles, integrating critical functionalities such as lane detection, traffic sign detection, vehicle detection and tracking, and vehicle-to-vehicle collaboration. We evaluate the performance of our autonomous systems compared to Tesla’s Full Self Driving (FSD), Waymo’s self-driving, and CommaAI’s openpilot system. Our autonomous driving system achieves 92% accuracy in lane detection and an accuracy of 94% for various traffic signs. The code to supplement this paper can be found at: https://github.com/mcropper14/Thesis-Code.