| 引用本文: | 徐君鹏,尹逊和.无线网络环境下移动机器人轨迹跟踪的认知控制分析与设计[J].控制理论与应用,2020,37(12):2511~2524.[点击复制] |
| XU Jun-peng,YIN Xun-he.Analysis and design of cognitive control for trajectory tracking of mobile robot in wireless network environment[J].Control Theory and Technology,2020,37(12):2511~2524.[点击复制] |
|
| 无线网络环境下移动机器人轨迹跟踪的认知控制分析与设计 |
| Analysis and design of cognitive control for trajectory tracking of mobile robot in wireless network environment |
| 摘要点击 1740 全文点击 567 投稿时间：2019-10-17 修订日期：2020-06-23 |
| 查看全文 查看/发表评论 下载PDF阅读器 |
| DOI编号 10.7641/CTA.2020.90869 |
| 2020,37(12):2511-2524 |
| 中文关键词 网络控制系统 模型预测控制 Q学习 认知控制 感知–作用循环 |
| 英文关键词 networked control systems model predictive control Q learning cognitive control perception-action cycle |
| 基金项目 中央高校基本科研业务费专项资金项目(2018YJS022)资助. |
|
| 中文摘要 |
| 现有的网络控制系统领域中的研究通常先对网络环境的时延和丢包特性进行理想性假设, 然后设计对应
的控制算法. 然而, 由于无线网络环境具有复杂的时延和丢包特性, 这些假设在无线网络控制系统(WNCS)运行过
程中很难得到满足. 为了在不对网络环境的时延和丢包特性进行理想性假设的前提下设计控制系统, 本文受认知
控制思想的启发, 在控制系统中加入了认知控制器. 认知控制器在感知–作用循环中学习产生认知作用的策略, 调
节无线网络的媒体接入控制(MAC)层的重传次数上限和物理控制器的命令序列长度, 使控制系统可以主动地适应
所处的无线网络环境. 本文以全向轮移动机器人为被控对象, 对使用认知控制器的WNCS和使用固定配置的WNCS
的仿真结果进行比较. 仿真结果表明使用认知控制器调节MAC层的重传次数上限和物理控制器的命令序列长度,
可以提高WNCS对网络环境的时延和丢包的承受能力. |
| 英文摘要 |
| Idealized assumptions about the character of delay and packet loss of network environment are first made in
existing research in the domain of networked control system, and then a corresponding control algorithm is designed based
on these assumptions. However, these assumptions hardly meet the realities during the operation of wireless networked
control system (WNCS) due to the complex character of delay and packet loss in wireless network environment. In order to
design the control system without making idealized assumptions of delay and packet loss in wireless network environment,
inspired by the idea of cognitive control, a cognitive controller is added into the control system. The cognitive controller
learns strategies of taking cognitive action in perception-action cycle, and the upper limit of the retry number of media
access control(MAC) layer in wireless network and the length of command sequence of the physical controller are adjusted
by the cognitive action of the cognitive controller. In this way, the control system can actively adapt to the wireless network
environment. In this paper, an omnidirectional wheel mobile robot is selected as the plant. The simulation results of WNCS
using cognitive controller and WNCS using fixed configuration are compared. Simulation results show that the cognitive
controller can improve the ability of WNCS to overcome delay and packet loss in the network environment by adjusting
the retry limit of the MAC layer and the length of command sequence of the physical controller. |