| 雾计算使能的移动机器人编队跟随研究与设计 |
| |
| 作者姓名： | 沈国锋 周明拓 李剑 王华俊 李凯 杨旸 |
| |
| 作者单位： | 1. 中国科学院上海微系统与信息技术研究所, 上海;2. 中国科学院大学, 北京 100049;3. 上海科技大学, 上海 201210 |
| |
| 基金项目： | 上海市科学技术委员会项目（18511106500）资助 |
| |
| 摘 要： | 多机器人系统中的跟随控制一直是研究热点，新兴的雾/边缘计算技术为机器人系统设计提供了新思路。本文提出无需全局定位信息的机器人编队控制方案，在系统实现上引入雾计算技术，卸载机器人计算任务。雾计算节点提供无线网络接入，运行跟随控制程序。跟随控制方案中，利用领航者航速信息计算期望跟随轨迹；通过视觉测量方法得到实时跟踪误差；采用航速重放和PD型迭代学习控制相结合的方法实现轨迹跟踪控制。原型实验表明，雾节点提供的网络可以满足实时控制需求，跟随控制程序的卸载降低了机器人本地算力要求，并节约了25%的计算能耗。3次场试中，编队平均跟踪误差在0.05 m以内，具有较好的精度。
|
| 关 键 词： | 雾计算 机器人编队 容器化 视觉测量 |
| 收稿时间： | 2019-10-09 |
| 修稿时间： | 2020-01-21 |
Research and design of fog computing-enabled mobile robots fleet tracking |
| |
| Authors: | SHEN Guofeng ZHOU Mingtuo LI Jian WANG Huajun LI Kai YANG Yang |
| |
| Institution: | 1. Shanghai Institute of Microsystem and Information Technology, Chinese Academy of Sciences, Shanghai;2. University of Chinese Academy of Sciences, Beijing 100049, China;3. ShanghaiTech University, Shanghai 201210, China |
| |
| Abstract: | Tracking control in multi-robot systems has always been a research hotspot, and the emerging fog/edge computing technology provides new ideas for the design of robot systems. A novel robot formation control scheme that requires no global positioning information is proposed in this article. Fog computing technology is also introduced to offload computing tasks when implementing the system. A fog node is responsible for providing wireless network and executing the tracking control program. As for the tracking control scheme, the desired tracking trajectory is calculated utilizing the speed information of the leader robot, and the real-time tracking error is obtained by vision based measurement. The trajectory tracking control is implemented by a combination of speed replay and PD-type iterative learning control method. The experimental results of our prototype system shows that the network provided by fog node can meet the needs of distributed real-time control applications. Local computing power requirement can be eased and 25% of computing energy is saved due to task offloading. The mean tracking error of the system is less than 0.05 m during field tests. |
| |
| Keywords: | fog computing robot formation containerization vision measuring |
|
| 点击此处可从《》浏览原始摘要信息 |
| 点击此处可从《》下载免费的PDF全文 |
