Joint Optimization of Cooperative Perception and Vehicle Control for V2V-enabled Autonomous Driving

Abstract

Cooperative perception (CP) is a pivotal technology for Autonomous Vehicles (AVs) to overcome the physical limitations of onboard sensors by sharing data via Vehicle-to-Vehicle (V2V) communication. However, the performance of CP is intrinsically coupled with vehicle control, as vehicle speed directly dictates the available time window for data transmission. Existing studies often treat these two aspects in isolation, neglecting their interaction. In this paper, we propose a Joint Cooperative Perception-Control Optimization (JCPCO) scheme to simultaneously optimize perception reliability and traffic efficiency. We formulate the joint decision-making problem as a multi-objective Mixed Integer Nonlinear Programming (MINLP) model that accounts for heterogeneous blind spot tasks, limited communication bandwidth, and safe mobility constraints. To address the NP-hard nature of the problem and meet the low-latency requirements of autonomous driving, we develop detailed linearization techniques and a computationally efficient heuristic algorithm, named JCPCO-Heuristic (JCPCO-H). Simulation results demonstrate that JCPCO-H achieves approximately 99.94% of the optimal system utility obtained by exact Mixed Integer Linear Programming (MILP) solvers while reducing computation time by over 98%. Furthermore, the proposed approach exhibits superior robustness and scalability compared to conventional baseline strategies across varying vehicle densities and task loads.