Enhancing Autonomous Vehicle Safety and Efficiency Using Deep Learning and V2V Communication

The world is rapidly advancing towards automation, and one of the core evolutions in automation in the development of Autonomous Vehicles, have the potential to revolutionize the transportation field. However, they are not deemed as safe as manual driving. Moreover, current AVs rely on numerous expensive sensors, increasing cost and complexity. These sensors also introduce limitations like blind spots and vulnerability to manipulation. In addition to this, depending solely on on-board sensors and preprogramed algorithms can hinder decision-making and collaboration with other vehicles in complex situations. This paper presents a comprehensive study on impact of combined power of Deep Learning and vehicle-to-vehicle communication to enhance the safety and efficiency of AVs while utilizing a reduced sensor suite. Deep learning algorithms optimize the use of these sensors, improving perception and prediction capabilities, potentially reducing the need for additional sensors and their associated costs. V2V communication facilitates real-time information exchange, allowing AVs to share critical data such as location, speed, and intent with surrounding vehicles. This collaborative approach provides a more comprehensive understanding of the environment, leading to informed decision-making, it could reduce rear-end crashes by up to 80% and lane-change crashes by up to 50% as compared to traditional sensor-heavy AV systems.