Parallel Distributed Cooperative Control of Multiple Vehicles with Low Communication Traffic

This paper presents a parallel distributed framework for cooperative control of of connected and automated vehicles (CAV) with low communication traffic. With this framework, the original centralized cooperative control problem is decomposed to a series of separable sub-optimization ones by introducing a set of consensus variables and linear transformation to fully use the computation resources of CAV. Additionally, to reduce the communication traffic between different computing nodes, a communication filtering strategy is designed by limit the transmission of the consensus variables by considering the fact that the interaction process tends to convergence when the variation among adjacent steps is small. To ensure the converge of numerical optimization process, a non-negative decreasing function is used to evaluate the variation of consensus variables. The effectiveness of the proposed cooperative control system is validated and analyzed by several comparative simulations with the original one by considering several indices including cooperation, optimality and communication efficiency.