Delay Mitigation for V2I-Based Cooperative Autonomous Driving Applications

Abstract

Perceiving the dynamic environment accurately is critical for safe intelligent driving. Vehicle-to-Infrastructure (V2I) communication is seen as one of the main enabling technologies for robust perception for autonomous vehicles, especially when the objects are heavily occluded or have small scales. However, the delay from edge computation and communication can significantly degrade the performance of existing cooperative perception methods. In this paper, we investigate the effects of delays in V2I-enabled autonomous driving applications. A configurable class-aware delay mitigation module is proposed to improve cooperative perception performance. By leveraging the object class information, the delay handling can provide predictive tracking, which improves the accuracy and reliability of the cooperative perception. Our experiments on the simulation platform show that the proposed approach has the potential to provide more accurate and robust object tracking for connected autonomous driving applications under communication delays.