The consensus problem of multi-agent systems (MASs) over directed graphs is analyzed in this paper. A sampled-data control law is induced from a nonlinear consensus protocol (continuous or not) by implementing feedback stabilization methods, in the sample-and-hold sense. Under the edge agreement framework, we show that consensus in a strongly connected network can be achieved over some suitable finite time. Furthermore, if an arbitrarily large and bounded actuator disturbance affects the MAS, we show that the sampled-data controller can be robustified to still guarantee the agreement. The case with observation errors is investigated too. The agent drift dynamics are required to be globally Lipschitz and bounded. Numerical simulations are finally proposed to validate the results.

本文分析了有向图上的多智能体系统（MAS）的共识问题。在采样保持的意义上，通过实施反馈稳定方法，可以从非线性共识协议（连续或不连续）中得出采样数据控制律。在边缘协议框架下，我们表明可以在某个合适的有限时间内达成强连接网络中的共识。此外，如果任意大且有界的执行器扰动影响了MAS，我们表明采样数据控制器可以得到增强以仍然保证一致性。观察误差的情况也被调查。代理漂移动力学要求全局为Lipschitz并有界。最后提出数值模拟以验证结果。