Offline optimization of sensor configuration and integration architecture for efficient cooperative navigation

Abstract

Multi-agent cooperative navigation can improve navigation performance by taking advantage of inter-agent communication and relative-sensing capabilities. For Multi-Agent Systems (MAS), the achievable navigation performance, communication cost, computation complexity, and sensor cost are directly influenced by the sensor configuration and the integration architecture. To reduce hardware cost and computation/communication load, this paper proposes a feasible method to realize the offline design of sensor configuration and integration architecture for cooperative navigation. Specifically, this goal is achieved by solving a multi-objective combinatorial optimization problem, where the sensor configuration and integration architecture are considered as the optimization variables; the navigation performance requirement is the constraint; and the communication cost, computation complexity, and sensor cost are considered as the optimization objectives. This optimization problem is solved by a Multi-Objective Simulated Annealing (MOSA) algorithm. Using a MAS with three unmanned aerial vehicles as an example, simulations are carried out to validate the proposed method, and the results show its feasibility and effectiveness. The Pareto solutions reveal the characteristics and applicability of different integration architectures in different scenarios, which is meaningful for practical applications.