RoCooper: Robust Cooperative Perception Under Vehicle-to-Vehicle Communication Impairments

Cooperative perception facilitated by vehicle-to-vehicle (V2V) data sharing has emerged as a crucial enabler for safe and efficient autonomous driving. However, the current state-of-the-art algorithms are unable to resolve severe performance degradation caused by communication impairments in realistic V2V scenarios. This paper models the V2V communication quality and confirms their fragile robustness under loss conditions. To this end, we propose RoCooper, a robust cooperative perception framework. It leverages lossless ego feature as an anchoring foundation, utilizing the multi-dimensional correlations of the features and dynamic regional selective cross-learning to perform multi-scale feature recovery and judiciously fuse multi-view features from neighboring vehicles. Specifically, in this process, we create three components: (i) Augmentor, which leverages historical information to perform spatio-temporal feature enhancement to preliminarily restore corrupted received features; (ii) Aggregator, which performs selective dynamic partitioning of features across multiple scales, and utilizes cross-learning and multi-scale merging to capture multi-granularity feature information; (iii) BlockPrioritizer, which dynamically assigns block weights and enables regional selective cross-learning, facilitating regional recovery and multi-view fusion at multiple scales. Extensive evaluations of real-world datasets demonstrate that our method achieves state-of-the-art performance in varying impairment scenarios.