Enhancing Safe Driving of Autonomous Buses in Obstructed-View Conditions Using Distributed Monocular Cameras

With the rapid advancement of autonomous driving technology, the application of automated urban transit has become a key component of modern transportation systems. As a representative of automated urban transit, the safe operation of autonomous buses is of great significance, especially in complex and variable traffic situations. This study proposes a low-cost method that combines a distributed architecture configuration based on monocular cameras with the latest YOLOv8 model to enhance the perception capabilities of fixed-route autonomous buses under specific conditions. The primary aim is to reduce traffic accidents caused by pedestrians suddenly appearing near the bus, particularly in blind spots. The methodology involves advanced object detection and transmission techniques. Video data captured by the monocular camera is processed in real-time, and data transmission is facilitated via Wi-Fi modules. The system centrally renders the relative positions of pedestrians and the bus on the receiving end, thereby enhancing the safety decision-making capabilities of the bus’s autonomous driving system. This approach not only advances the development of automated urban transit systems but also promotes the practical implementation of low-cost autonomous driving solutions, particularly in scenarios where the view is obstructed for autonomous buses.