A distributed multi-objective optimization algorithm with time-varying priorities for multi-agent systems

Abstract Real-world multi-agent systems often operate in dynamic environments where the importance of tasks evolves over time. Traditional multi-objective optimization methods typically assume fixed or uniformly weighted objectives, limiting the system’s adaptability and responsiveness. This paper presents a new distributed multi-objective optimization (MOO) algorithm designed for multi-agent systems (MASs) with time-varying priorities. Building upon previous foundational work, the proposed algorithm addresses the dynamic nature of real-world applications by allowing agents to adjust their priorities over time. This flexibility enhances the system’s adaptability and responsiveness, which is crucial for emergency responses and complex robotic coordination tasks. Mathematical proofs of convergence and performance bounds are provided, confirming the algorithm’s effectiveness in managing conflicting objectives in dynamic environments. The algorithm’s performance is demonstrated through simulations involving small and large-scale agent networks. To further demonstrate the algorithm’s applicability in a simulated environment, it has been employed to generate the reference trajectory for a swarm of autonomous ground robots in a simulated rendezvous problem. Additionally, a Sugeno-type fuzzy controller has been designed to enhance the robots’ trajectory-following capabilities within the simulation. The results of the simulation of the swarm of robots for the application of the rendezvous problem with the fuzzy controller are presented proving the performance of both the algorithm and the controller.