CASIA OpenIR  > 毕业生  > 硕士学位论文
基于云服务的娱乐服务机器人识别算法与系统架构研究
Alternative TitleRecognition Algorithm and System Architecture of Entertainment Service Robot Based on Cloud Services
向廷
Subtype工程硕士
Thesis Advisor李成荣
2015-05-29
Degree Grantor中国科学院大学
Place of Conferral中国科学院自动化研究所
Degree Discipline控制理论与控制工程
Keyword娱乐服务机器人 系统架构 云服务 模糊c均值聚类 最大期望算法 Entertainment Service Robot System Architecture Cloud Service Fuzzy C-means Algorithm Em Algorithm
Abstract随着计算机技术、传感器技术、人工智能等领域突飞猛进的发展,机器人逐渐摆脱它原有的古板笨拙且专有性极强的制造业和军事工业领域的束缚,开始向民用服务领域大举迈进。同时近年来移动互联网的爆炸式发展,给娱乐服务机器人注入了更大的想象力和灵活的空间,应用领域得到了极大的扩展延伸。 本文以娱乐服务机器人大规模应用的蓝图为宗旨,提出了一种着眼于未来娱乐服务机器人应用的三层系统架构,研究了以魔方机器人、家庭服务机器人及售货机器人原型为基础的分层功能实现。论文的主要工作包括: 1、提出了一种由机器人本体、云服务后台、远程交互端组成的娱乐服务机器人三层系统架构,以实现机器人本体系统结构的集约化、数据服务平台和计算服务平台的系统化及远程交互流程的统一化。 2、在机器人主体层,以魔方机器人为例探讨了其色块颜色的识别问题,提出了一种类空间域固定的模糊C均值聚类算法。由于魔方状态空间的限定性,适合于用聚类的方法完成颜色识别,论文对多种聚类算法进行了对比和分析。 3、提出了一种混合光源环境下魔方颜色识别的方法。由于在混合光源环境下,单一色块的颜色空间分布近似服从高斯混合模型,因此采用EM算法进行预处理提取其中的各模型分量,进而结合桶分类的思想构造了一种三层级联算法。 4、在云服务后台层,以魔方机器人为例论述了解魔方算法Kociemba++在Spark计算平台的部署,以家庭服务机器人为例论述了基于ICTCLAS的语义理解和基于Face++的人脸识别在SAE数据平台上的应用。在远程交互层,以售货机器人为例实现了售货机器人综合信息的交互与显示。
Other AbstractAlong with the development of computer science, sensor technology and artificial intelligence by leaps and bounds, the robot has stepped out of its original and proprietary fields such as manufacturing and military industries gradually, and begins to step into the civil service field aggressively. Moreover, with the explosive growth of mobile Internet in recent years, the entertainment service robot have been injected into great imagination and flexible space, application fields are expanding constantly. Based on the blueprint of entertainment service robots’ large-scale application, this paper proposes a three layer system architecture which is aimed at the future, and studies on the function and implementation of each layer with the Rubik’s cube robot, family service robot and vending robot. The main works of this paper are as follow: Firstly, a system architecture of entertainment service robot that composed of robot’s main body, cloud services platform and remote interaction entrance is proposed, which is aimed to achieve intensive robot ontology system structure, systematic data exchange and processing platform and unified remote interaction process. Secondly, in the robot’s main body layer, color recognition of the Rubik's cube is discussed and a kind of fixed spatial domain fuzzy c-means clustering algorithm is put forward. Due to the restriction of the Rubik's cube’s state space, this problem is suitable for clustering algorithm to solve. In this section, a variety of clustering algorithms are compared. Thirdly, a method which can solve the color recognition of the Rubik’s cube robot under the mixed lighting environment is proposed. Considering that single color cube’s color space distribution under the hybrid lighting environment is closed to the Gauss mixture model, EM algorithm is adopted to extract the components, and then combine with the bucket classification to form a three layer cascade algorithm. Finally, in the cloud service layer, the deployment of Kociemba++ algorithm on spark computing platform for the Rubik’s cube robot and the implement of face recognition and semantic understanding on SAE data and service platform for the home service robot are discussed; In the remote interaction layer, integrated information’s interaction is showed.
Other Identifier2012E8014661088
Language中文
Document Type学位论文
Identifierhttp://ir.ia.ac.cn/handle/173211/7745
Collection毕业生_硕士学位论文
Recommended Citation
GB/T 7714
向廷. 基于云服务的娱乐服务机器人识别算法与系统架构研究[D]. 中国科学院自动化研究所. 中国科学院大学,2015.
Files in This Item:
File Name/Size DocType Version Access License
CASIA_2012E801466108(2335KB) 暂不开放CC BY-NC-SAApplication Full Text
Related Services
Recommend this item
Bookmark
Usage statistics
Export to Endnote
Google Scholar
Similar articles in Google Scholar
[向廷]'s Articles
Baidu academic
Similar articles in Baidu academic
[向廷]'s Articles
Bing Scholar
Similar articles in Bing Scholar
[向廷]'s Articles
Terms of Use
No data!
Social Bookmark/Share
All comments (0)
No comment.
 

Items in the repository are protected by copyright, with all rights reserved, unless otherwise indicated.