科学技术的发展使机器人成为人类社会的组成部分，多样的机器人服务对高计算力的需求与其有限的机载资源之间的矛盾日益凸显。云机器人系统的提出使机器人可通过云计算获得云端无限的计算和存储资源，提升自身服务能力，而计算卸载作为云机器人的重要特征是实现任务迁移，解决上述矛盾的有效手段。现今的云机器人系统研究多集中于云端服务系统设计，而爆发增长的边缘设备数据使该模式下的中心云面临着服务响应不及时、数据丢失、带宽压力大等通信挑战；其次在复杂多变的无线网络环境中，如何保证机器人计算任务的稳定卸载是云机器人系统中进行计算卸载面临的决策挑战；最后如何在已有计算卸载决策下，设计与实现适用于群组机器人的计算卸载原型系统是云机器人系统发展亟需解决的需求。针对上述问题，本文结合云机器人系统架构分别从计算卸载决策算法和可行性系统设计角度进行研究。本文主要工作和创新成果如下：

首先引入边缘计算概念，根据任务卸载交互方式对云机器人系统中的任务计算卸载问题进行分析，结合时延和能耗评价指标分别建立了任务在本地和边缘侧执行的成本函数。其次将机器人与边缘云服务器间的竞争过程建模为Stackelberg博弈模型，利用逆向归纳法将博弈过程分解为二阶段博弈问题，求解了机器人的最佳卸载策略解和边缘云的最佳价格策略解。最后设计了基于汉明距离的动态规划算法对计算任务最终卸载决策结果进行求解。仿真实验结果表明，本文提出的方法能有效降低机器人的计算成本、减少任务执行总时间和能量消耗，同时使边缘云在有限资源限制下实现收益最大化。

针对云机器人系统的可行性原型系统研究，本文在计算卸载决策结果的基础上，以机器人同时定位与地图构建（SLAM）任务为例，分析了在资源受限机器人上部署SLAM任务过程中面临的问题，通过改进ORB-SLAM算法，将部分计算任务迁移到边缘云中执行，利用边缘云中资源实现计算任务的加速，并对改进算法的性能及可能的计算卸载策略进行了分析。之后结合相关文献和软件设计了云机器人的计算卸载原型系统，在最大程度满足机器人任务执行稳定性与实时性的条件下，充分利用边端协同计算优势，节约本地计算和内存资源开销。通过实验证明了上述原型系统的可行性和高效性，同时证明了该方法对提升机器人系统执行效率和服务质量，降低机器人硬件要求发挥了积极作用。

With the development of science and technology, robots have become an integral part of human society, the contradiction between the demand for high computing resources of various robot services and limited onboard resources has become increasingly prominent. The concept of cloud robotics system enables the robot to obtain unlimited computing and storage resources in the cloud through cloud computing, also can improve the service ability of robots. Computing offloading is an important feature of cloud robotics, it is an effective means to realize tasks offloading and slove above contradiction. Nowadays, the research of cloud robotics system focus on the design of cloud service system, with the explosive growth of data in edge device makes the central cloud have to face with communication challenges such as time delay in service response, data loss, pressure on bandwidth and so on; Secondly, wireless network environment can be complex and changeable, how to ensure the stable offloading of robot computing tasks is a decision-making challenge for computing offloading in cloud robotics system; Finally, how to design and implement a computing offloading prototype system suitable for group robots under the existing computing offloading decisions, it is an urgent need to be solved in the development of cloud robot system. To solve the above problems, combined with the architecture of cloud robotics, this paper studies computing offloading decision algorithm and feasibility system design. The main work and innovative achievements of this paper are as follows:

(1) Firstly, the concept of edge computing is introduced into cloud robotics, this paper analyzed the problem of computing offloading in cloud robotics system according to the interactive mode of task offloading. The cost functions of task execution on the local and edge sides has established respectively with the evaluation indexes of time delay and energy consumption. Secondly, the competition process between the robot and the edge cloud server is modeled as a Stackelberg game model, the game process is decomposed into two-stage game problems by using the reverse induction method. The optimal offloading strategy solution of the robot and the optimal price strategy solution of the edge cloud are solved. Finally, a dynamic programming algorithm based on Hamming distance is designed to solve the final offloading decision result of the computing task. The simulation results show that the proposed method can reduce the computing cost of the robot effectively, reduce the total task execution time and energy consumption, and maximize the revenue of the edge cloud under the limitation of limited resources.

(2) For the feasibility prototype system research of cloud robotics system, based on the result of tasks offloading decision, taking the Simultaneous Localization and Mapping on robot as an example, this paper analyzes the problems appeared in the process of deploying SLAM task on the resource limited robot. By the improving the ORB-SLAM algorithm, part of the computing tasks are migrated to the edge cloud for execution, and the resources in the edge cloud are used to accelerate the computing tasks. The performance of the improved algorithm and the possible computing offloading strategy are analyzed. Combined with relevant literature and software, the prototype system of computing offloading of robotics is designed. Under the condition of stability and real-time performance of robot task execution to the greatest extent, the advantages of edge collaborative computing are fully used to save the cost of local computing and memory resources. The feasibility and efficiency of the above prototype system are proved by experiments, and it also plays a positive role in improving the overall efficiency and service quality of the robot system and reducing the hardware requirements of the robot.

[1] 施俊屹,查富生,孙立宁,郭伟,王鹏飞,李满天.移动机器人视觉惯性SLAM研究进展[J].机器人,2020,42(06):734-748.

[2] 张松. 基于云计算的服务机器人系统设计与实现[D].西安科技大学, 2020.

[3] Kuffner J. Cloud-enabled humanoid robots[C]//Humanoid Robots (Humanoids), 2010 10th IEEE-RAS International Conference on, Nashville TN, United States, Dec. 2010.

[4] 陈玉平,刘波,林伟伟,程慧雯.云边协同综述[J].计算机科学,2021,48(03):259-268.

[5] 谢人超,廉晓飞,贾庆民,黄韬,刘韵洁.移动边缘计算卸载技术综述[J].通信学报,2018,39(11):138-155.

[6] Arumugam R, Enti V R, Bingbing L, et al. DAvinCi: A cloud computing framework for service robots[C]//2010 IEEE international conference on robotics and automation. IEEE, 2010: 3084-3089.

[7] Waibel M, Beetz M, Civera J, et al. Roboearth[J]. IEEE Robotics & Automation Magazine, 2011, 18(2): 69-82.

[8] Hunziker D, Gajamohan M, Waibel M, et al. Rapyuta: The roboearth cloud en-gine[C]//2013 IEEE international conference on robotics and automation. IEEE, 2013: 438-444.

[9] Kehoe B, Matsukawa A, Candido S, et al. Cloud-based robot grasping with the google ob-ject recognition engine[C]//2013 IEEE International Conference on Robotics and Automa-tion. IEEE, 2013: 4263-4270.

[10] Sorrentino A, Cavallo F, Fiorini L. A plug and play transparent communication layer for cloud robotics architectures[J]. Robotics, 2020, 9(1): 17.

[11] Chen K, Lu D, Chen Y, et al. The intelligent techniques in robot kejia–the champion of ro-bocup@ home 2014[C]//Robot Soccer World Cup. Springer, Cham, 2014: 130-141.

[12] 蔡增玉,王文倩,赵振宇,张建伟,冯媛.基于H.265的云机器人图像采集系统设计与实现[J].现代电子技术,2021,44(05):66-69.

[13] Xie Y, Guo Y, Mi Z, et al. Loosely coupled cloud robotic framework for qos-driven re-source allocation-based web service composition[J]. IEEE Systems Journal, 2019, 14(1): 1245-1256.

[14] 董思岐,李海龙,屈毓锛,张钊,胡磊.移动边缘计算中的计算卸载策略研究综述[J].计算机科学,2019,46(11):32-40.

[15] Zhao P, Tian H, Qin C, et al. Energy-saving offloading by jointly allocating radio and computational resources for mobile edge computing[J]. IEEE Access, 2017, 5: 11255-11268.

[16] Yu H, Wang Q, Guo S. Energy-efficient task offloading and resource scheduling for mobile edge computing[C]//2018 IEEE international conference on networking, architecture and storage (NAS). IEEE, 2018: 1-4.

[17] Xu F, Yang W, Li H. Computation offloading algorithm for cloud robot based on improved game theory[J]. Computers & Electrical Engineering, 2020, 87: 106764.

[18] Tao X, Ota K, Dong M, et al. Performance guaranteed computation offloading for mo-bile-edge cloud computing[J]. IEEE Wireless Communications Letters, 2017, 6(6): 774-777.

[19] Liu J, Mao Y, Zhang J, et al. Delay-optimal computation task scheduling for mobile-edge computing systems[C]//2016 IEEE international symposium on information theory (ISIT). IEEE, 2016: 1451-1455.

[20] Mao Y, Zhang J, Song S H, et al. Power-delay tradeoff in multi-user mobile-edge compu-ting systems[C]//2016 IEEE Global Communications Conference (GLOBECOM). IEEE, 2016: 1-6.

[21] Kamoun M, Labidi W, Sarkiss M. Joint resource allocation and offloading strategies in cloud enabled cellular networks[C]//2015 IEEE International Conference on Communica-tions (ICC). IEEE, 2015: 5529-5534.

[22] Zhang H, Guo J, Yang L, et al. Computation offloading considering fronthaul and backhaul in small-cell networks integrated with MEC[C]//2017 IEEE Conference on Computer Communications Workshops (INFOCOM WKSHPS). IEEE, 2017: 115-120.

[23] You C, Huang K. Multiuser resource allocation for mobile-edge computation offload-ing[C]//2016 IEEE Global Communications Conference (GLOBECOM). IEEE, 2016: 1-6.

[24] Chen M H, Liang B, Dong M. Joint offloading decision and resource allocation for mul-ti-user multi-task mobile cloud[C]//2016 IEEE International Conference on Communica-tions (ICC). IEEE, 2016: 1-6.

[25] Muñoz O, Pascual-Iserte A, Vidal J. Joint allocation of radio and computational resources in wireless application offloading[C]//2013 Future Network & Mobile Summit. IEEE, 2013: 1-10.

[26] Munoz O, Pascual-Iserte A, Vidal J. Optimization of radio and computational resources for energy efficiency in latency-constrained application offloading[J]. IEEE Transactions on Vehicular Technology, 2014, 64(10): 4738-4755.

[27] Zhang K, Mao Y, Leng S, et al. Optimal delay constrained offloading for vehicular edge computing networks[C]//2017 IEEE International Conference on Communications (ICC). IEEE, 2017: 1-6.

[28] 危泽华,曾玲玲.基于Stackelberg博弈论的边缘计算卸载决策方法[J].数学的实践与认识,2019,49(11):91-100.

[29] Ma X, Lin C, Xiang X, et al. Game-theoretic analysis of computation offloading for cloud-let-based mobile cloud computing[C]//Proceedings of the 18th ACM International Confer-ence on Modeling, Analysis and Simulation of Wireless and Mobile Systems. 2015: 271-278.

[30] Zhang J, Xia W, Yan F, et al. Joint computation offloading and resource allocation optimi-zation in heterogeneous networks with mobile edge computing[J]. IEEE Access, 2018, 6: 19324-19337.

[31] Jie Y, Tang X, Choo K K R, et al. Online task scheduling for edge computing based on re-peated Stackelberg game[J]. Journal of Parallel and Distributed Computing, 2018, 122: 159-172.

[32] 贾万睿. 基于移动边缘计算的任务协同卸载策略[D].电子科技大学, 2020.

[33] 李波, 薛端, 黄鑫. 云机器人系统研究综述[J]. 计算机工程与应用, 2017, 53(17):26-40.

[34] 蔡鑫伟,侯向辉,莫清宇,张美玉,简琤峰.面向群组机器人路径规划的Voronoi-APF算法研究[J].小型微型计算机系统,2021,42(12):2490-2495.

[35] 王伟嘉,郑雅婷,林国政,张亮,韩战钢.集群机器人研究综述[J].机器人,2020,42(02):232-256.

[36] 王常虹,刘博,李清华.集群多机器人系统建模研究的发展与展望[J].导航定位与授时,2021,8(01):1-13.

[37] 杨斯博,李敏强.一个基于自组织多Agent系统的智能控制与决策模型[J].天津大学学报,2012,45(10):903-911.

[38] 武文亮,周兴社,沈博,赵月.集群机器人系统特性评价研究综述[J/OL].自动化学报:1-24[2022-05-26].

[39] 洪臣,史殿习.云机器人架构和特征概述[J].机器人技术与应用,2018(06):15-19.

[40] Furrer J, Kamei K, Sharma C, et al. UNR-PF: An open-source platform for cloud net-worked robotic services[C]//2012 IEEE/SICE International Symposium on System Integra-tion (SII). IEEE, 2012: 945-950.

[41] Hu Y C, Patel M, Sabella D, et al. Mobile edge computing—A key technology towards 5G[J]. ETSI white paper, 2015, 11(11): 1-16.

[42] 王大伟,王卓,王鹏,王志刚,王海涛.基于边缘计算的云原生机器人系统[J].智能科学与技术学报,2020,2(03):275-283.

[43] 陈友东,胡嘉航.基于边缘计算的工业应用：自动导引小车控制系统[J].计算机集成制造系统,2019,25(12):3191-3198.

[44] Chen Y, Feng Q, Shi W. An industrial robot system based on edge computing: An early experience[C]//USENIX Workshop on Hot Topics in Edge Computing (HotEdge 18). 2018.

[45] 郭宇.云机器人系统网络优化与计算卸载相关技术研究[D].北京科技大学,2020.

[46] 杨卫霞.基于博弈论的云机器人计算卸载策略研究[D].西安工业大学, 2020.

[47] 盛津芳,滕潇雨,李伟民,王斌.移动边缘计算中基于改进拍卖模型的计算卸载策略[J].计算机应用研究,2020,37(06):1688-1692.

[48] 张继鑫,武延军.基于ROS的服务机器人云端协同计算框架[J].计算机系统应用,2016,25(09):85-91.

[49] Marden J R, Shamma J S. Game theory and distributed control[M]//Handbook of game theory with economic applications. Elsevier, 2015, 4: 861-899.

[50] Churkin A, Bialek J, Pozo D, et al. Review of Cooperative Game Theory applications in power system expansion planning[J]. Renewable and Sustainable Energy Reviews, 2021, 145: 111056.

[51] 孙弋,许志杰.云机器人系统的计算卸载策略研究[J].西安理工大学学报,2021,37(04):562-569.