Realistic Implementation of Cloud Scheduling Algorithm ICU in the Floor-climbing Robet with High Efficiency

• Main Authors: Yu-Shiu Cheng, 鄭羽修
• Other Authors: Ching-Kuo Wang
• Format: Others
• Language: zh-TW
• Published: 2015
• Online Access: http://ndltd.ncl.edu.tw/handle/76052632961129678815

碩士 === 華夏科技大學 === 智慧型機器人研究所 === 104 === In recent years, the rapid development of flexible manufacturing system (FMS) with Industry 4.0 in German and Intelligent Manufacturing 2025 in China, the intelligent robotics will become a promising technology of the 21st century. In the realistic implementation of cloud computing technology, virtualization (Virtualization) of artificial intelligence is a resource management technology, is a variety of physical resources of the computer, such as server, network, memory and storage devices, to be abstract, after the conversion revealed, inter breaking entity structure non-cutting obstacles for the effective communication of the multi-robot systems. This research is dedicated to construct a virtualized environment, which allows different physical stations of mudulized sub-systems to perform multiplex computation in order to have better operating speed, lower cost efficiency, more robust integration of data grasping capability. On the other hand, it also has been made to reach a better virtualized resource management of epoch-making significance, adjust virtual machine in the Intelligent Control Unit (ICU) in the proposed intelligent robotic system as a key parameter. The proposed algorithm will directly affect how the virtual machine overall system performance in realistic implementation of the virtualized cloud computing system. In the end of this research, the proposed credit scheduler XenServer successfully modulates ICU time on a fair and efficient for each virtual ICU, but encounters a parallel situation, the performance will subsequently dropped, via the proposed experimental analysis XenServer's ICU wisdom adjustment algorithms to achieve without affecting the additional resource of parallel execution performance. It is also proved to be a better choice of the program and then parallel performance in reaching the goal of this master's thesis research two years.