基于演化博弈的多智能体环形编队的理论与实验研究

Formation control is one of the key technology for multi-robot cooperation which has an important application value. As a benchmark problem in formation control, circle formation problem has attracted extensive attention from researchers. Evolutionary game theory is a powerful tool for the investigation of how cooperation might evolve in a population of self-regarding agents. This proposal introduces the evolutionary game theory to the circle formation problem for multi-agent systems. A coevolutionary dynamics model of multi-agent systems combined with evolutionary game theory will be proposed, and the coevolution of the dynamics of cooperation in the population and the dynamics of system will be investigated. We will decide some decisive factors and mechanisms which affect the level of cooperation, and analyze how they affect the system dynamics. Some reasonable performance indices are proposed and the mathematical description and model of the performance optimization problem will be given. Then, we will design control protocols and the strategy update rules in our proposed coevolutionary dynamics model of multi-agent systems based on evolutionary game theory to solve the optimization problem. Finally, an experimental platform for the circle formation problems for multiple robotic fish will be constructed, and we will conduct experiments to verify our theoretical results. The research results of this proposal will provide strong support to perfect the research on formation problems of multi-agent systems and provide new idea and method for the analysis and control of networked multi-agent systems.

多机器人协作技术具有巨大的应用价值，编队控制是多机器人协作应用的核心支撑技术。环形编队是其中的标准问题，受到国内外学者的广泛关注。演化博弈是研究自私个体如何产生合作行为的最有力工具，本项目首先将演化博弈理论引入到多智能体系统环形编队控制问题中，提出一个融合演化博弈理论的多智能体系统动力学模型；然后研究群体合作动力学和多智能体系统动力学的共演化的问题，确定若干影响合作水平的决定性的因素和机制，定量分析其对系统动力学的影响；接着提出合理性能优化指标，设计控制协议和演化规则，对系统进行性能优化；最后，展开多机器鱼环形编队的实验研究，对所得理论进行验证和评价。本项目成果将有助于从理论上拓展多智能体系统协作控制的研究，为多智能体系统的分析与控制提供新思路和新方法。

多机器人协作技术具有巨大的应用价值，编队控制是多机器人协作应用的核心支撑技术。环形编队是其中的标准问题，受到国内外学者的广泛关注。另一方面，演化博弈是研究自私个体如何产生合作行为的最有力工具。本项目首先针对经典的多智能体系统，研究编队控制等协调控制问题，提出基于极限环的控制协议设计方法，所提出的控制律形式简洁、可在个体的随体坐标系下应用，在所提出的控制律下，系统收敛到期望的编队且个体之间不发生碰撞。然后，基于上述多智能体系统的研究，引入演化博弈理论，提出一个基于演化博弈的多智能体系统动力学模型，研究群体合作动力学和系统动力学的共演化和系统性能优化的算法设计。最后，展开多机器鱼环形编队的仿真和实验研究，对所得理论进行验证和评价。本项目成果将有助于从理论上拓展多智能体系统协作控制的研究，为多智能体系统的分析与控制提供新思路和新方法。