Cloud-Assisted 360-Degree 3D Perception for Autonomous Vehicles Using V2X Communication and Hybrid Computing

A key challenge for autonomous driving lies in maintaining real-time situational awareness, particularly in complex urban settings. This study introduces an innovative cloud-driven solution for 360-degree perception in autonomous vehicles using vehicle-to-everything (V2X) communications. Our approach utilizes transformer-based models to fuse multi-camera sensor data into a comprehensive bird’s-eye view (BEV) representation, enabling accurate 3D object detection. By offloading computationally intensive tasks to the cloud, the system achieves scalable processing while reducing latency. In addition, techniques such as feature vector clipping, compression, and quantization are applied to optimize data transmission, ensuring real-time performance. Experimental results show an over 79.2% reduction in end-to-end delay compared to onboard-only computing. These experiments validate the integration of AI and V2X technology to enhance autonomous vehicle perception.