确保网络化多机器人协同跟踪的实时通信条件研究

In a multi-agent cooperative control system, the communication between agents plays a critical role in the cooperation of multiple agents. This project considers a typical example of such systems, a cooperative tracking system of multiple robots, and studies the cooperative tracking problem where the state information is transmitted over digital networks. In such a system, the state transmission network is composed of multiple communication paths, whose real-time communication properties include communication bandwidth, dropout condition (the dropout rate, dropout pattern and dropout acknowledgement) and communication delay. These individual paths together determine the real-time communication properties of the overall state transmission network. This project investigates the following four problems. 1. What real-time communication condition should the overall network satisfy in order to achieve the cooperative tracking goal? Here the concerned multiple robots could be heterogeneous. 2. Under a given real-time network condition, design quantization and local control algorithms for these robots so that the desired overall cooperative tracking performance can be guaranteed. 3. Design a control event based communication policy, under which the transmission of states of robots are triggered by some control variables and the total amount of transmitted data could be reduced without significantly degrading the overall cooperative tracking performance. 4. Develop simulation and experimental plantforms to verify the obtained theoretical results. The major goal of this project is to establish a theorectical framework to quantitatively describe the relationship among the real-time network communication conditions, the dynamics of robots and the cooperative tracking performance.

多自主体协同控制系统中，自主体间的通信对实现协同目的十分重要。本项目将该类系统的典型代表- - 多机器人协同跟踪系统作为对象，研究基于数字网络实现状态信息传输时的协同跟踪问题。状态传输网络由多条通信信道组成，每条信道的实时通信特性由其有限的通信带宽、丢包特性（丢包率、丢包模式及丢包确认）和时延特性来表征，这些信道的特性综合起来决定了整体传输网络的实时通信特性。本项目研究下列问题：1、整体网络的实时通信特性满足什么条件，才能实现全局协同跟踪（机器人间可能存在动态特性差异）；2、在给定网络条件下，设计出能确保全局跟踪性能的多机器人的状态量化算法与局部控制算法；3、设计控制事件驱动的状态通信机制，即通过控制变量事件触发状态信息传输、在不损害系统跟踪性能的前提下降低通信数据量；4、开发仿真程序与实验平台来验证理论结果。本项目力求建立网络实时通信条件、机器人动态特性、协同跟踪性能相互间关系的理论框架。

本项目研究了基于数字网络实现状态信息传输的多机器人协同跟踪问题。作为一种典型多自主体系统，机器人之间的通信对于实现系统协同控制目的十分重要。在本项目中，机器人的状态传输网络由多条通信信道组成，每条信道的实时通信特性由其有限的通信带宽、丢包特性（丢包率、丢包模式及丢包确认）和时延特性来表征，这些信道的特性综合起来决定了整体传输网络的实时通信特性。本项目取得了如下成果：1、指出了整体网络的带宽满足什么条件，才能实现全局协同跟踪；设计了合适的状态量化算法和机器人协调控制算法来实现协同跟踪目的，并且所设计的算法具有较高的鲁棒性，即尽量少地依赖机器人通信拓扑的全局信息。2、提出了一种迭代式的多机器人局部控制增益设计方法，能够实现全局协同控制性能的优化。3、将事件触发引入网路化控制系统采样中，即状态变量基于特定事件的触发进行传输，在不损害系统跟踪性能的前提下降低了通信数据量，突破了确保系统稳定性的传统反馈带宽极限；尤其是该事件触发策略充分考虑了未知计算延时、未知网络传输延时、可能的丢包对于确保系统稳定性的带宽条件的影响。4、开发了基于MATLAB的仿真程序，开发了基于ARM11的实验平台来对理论结果进行了验证。本项目建立le 网络通信条件、机器人动态特性、协同跟踪性能相互间关系的理论框架。