智能群体通常是指由大量个体经局部动态相互作用而涌现整体有序行为的系统。复杂网络理论为描述群体内部个体之间的相互作用关系提供了有效框架。从控制科学与工程的角度而言，采用演化博弈理论对网络化群体行为进行建模分析与协作控制的研究亟待展开。针对此热点问题，本项目将融合分布式优化的思想，针对局部信息不完备的情况，建立基于概率切换的演化动力学模型。计划结合系统控制理论和演化博弈理论，分析系统演化的稳定性、收敛性及其他动力学特征。在理论分析的基础上，进而开展多机器人系统协作控制算法设计与实验研究，实现对网络化智能群体的有效干预与控制。建立的多机器人系统协作控制平台和算法评价方案，可为多个体分布式工程应用提供必要的理论基础与方法。

A smart swarm consists of a large number of autonomous agents that behave intelligently at the group level. Complex network theory has provided an effective framework to model the interactions among agents in a group. From the perspective of control theory and control engineering, the dynamical system modeling based on evolutionary game theory, cooperative control theory and methodology for swarms are in need of investigation. Focusing on this hot topic, in the proposed project we plan to establish a probabilistic switching model of evolutionary dynamics under the condition of the imperfect information, combined with the ideas from the study of distributed optimization. We aim at gaining the theoretical results about stability, convergence and other key dynamic characteristics of the system. Further, the physical experiments using multiple robots will be established, and some hints about the effective intervention and control for the networked swarm are thus expected. The established robotic platform for the cooperation control of a multi-agent system, and the algorithm evaluation scheme, can be useful by providing new theory and methodology for various agent-based distributed engineering applications.