Scalable and Dynamic Cooperative Perception: A Data/Model Co-Driven Framework

Cooperative perception (CP) is a key approach to ensuring reliable situation awareness of connected and autonomous vehicles (CAVs). In this article, we discuss the key challenges in terms of scalability, dynamics, and performance uncertainty for supporting CP in a practical network environment. Then, we present a data/model co-driven framework for scalable and dynamic CP with performance awareness, as an engineering solution to address the challenges. Specifically, we propose a performance-aware scalable CP scheme based on a learning-assisted optimization approach and a dynamic CP scheme based on an optimization-assisted learning approach for different scenarios, both exploiting data-driven and model-based methods to enhance each other. Finally, a case study is presented to show the effectiveness of our scheme in handling the network dynamics with resource efficiency.