针对未知环境下的群体机器人形成聚集动态队形的问题,提出k-means聚类算法,将该算法与基于行为的控制相结合,使群体机器人形成聚集队形,减少机器人在聚集位置的抖动问题。引入四种适应度函数,提高机器人对未知环境的反应速度。个体机器人可根据局部感知自主产生队形控制向量。仿真实验表明了队形控制策略的可行性与有效性。 Aiming at the formation of aggregated dynamic formation of group robots in unknown environment, a k-means clustering algorithm is proposed. Combining this algorithm with behavior-based control, the group robots form an aggregation formation and reduce the jitter of the robots in the aggregation position . Four fitness functions are introduced to improve the robot’s response speed to unknown environment. Individual robots can generate formation control vectors autonomously according to local perception. Simulation results show that the formation control strategy is feasible and effective.