This paper investigates the consensus problem for nonlinear heterogeneous multi-agent systems with limited communication data rate. Each agent is modeled by a higher-order strict-feedback continuous-time system with unknown nonlinearities and external disturbance, and only the first state variable being measurable. Extended state observers (ESOs) are used to estimate the unmeasurable agent states and the unknown nonlinear dynamics. An ESO-based distributed output feedback protocol with dynamic encoding and decoding is then presented. It is shown that, for a connected undirected network, the proposed protocol guarantees practical output consensus, in which the steady-state consensus error can be made arbitrarily small. The ESO-based protocol also shapes the transient consensus performance, as it is capable of recovering the consensus performance of a linear counterpart with fully measurable states. Furthermore, we prove that for higher-order uncertain nonlinear multi-agent systems, consensus can be achieved with merely one bit information exchange between each pair of adjacent agents at each time step. Finally, simulations on third-order pendulum systems are given, which verify the theoretical results.

本文研究了具有有限通信数据速率的非线性异构多智能体系统的共识问题。每个代理都由具有未知非线性和外部干扰的高阶严格反馈连续时间系统建模，并且只有第一个状态变量是可测量的。扩展状态观察器（ESO）用于估计不可测量的主体状态和未知的非线性动力学。然后提出了一种基于ESO的具有动态编码和解码的分布式输出反馈协议。结果表明，对于连接的无向网络，所提出的协议保证了实际的输出一致性，其中稳态一致性误差可以任意减小。基于ESO的协议还可以塑造瞬态共识性能，因为它能够恢复具有完全可测状态的线性对等方的共识性能。此外，我们证明了对于高阶不确定非线性多智能体系统，在每个时间步长上，每对相邻智能体之间仅进行一位信息交换就可以达成共识。最后，对三阶模拟给出了摆系统，验证了理论结果。