Research And Implementation On Cloud Application Architecture And Management Framework For Distributed Robotics

With the development of robotics,cloud robots and distributed(cluster)robots are gradually playing an important role in the industry and people's life.Cloud robots can improve the computing and storage ability through the cloud server which has high performance.And with the collaboration of robots,distributed robots can improve the efficiency greatly.Based on the combination of cloud computing and distributed robots,it is becoming a hot topic in academia and industry to use the cloud server to realize the collaborative planning and capability improvement of robots by using various kinds of information.At present,Researches focused on "cloud + distributed robot" field are mostly aimed at a specific application scenario and specific robot types,and don not have versatility.Developers should not only pay attention to the algorithm,but also give high priority to the management and control of the distributed robot when developing applications for robots.Besides the development work,developers also need to deploy applications to the cloud server and robots as well as managing and control these applications.In order to solve these problems,we focus on the implementation of management and control on robot and applications in order to reduce the cloud application developers' workload,so that it does not need to care about the management and control of the robots and its applications.In this paper,ROS is selected as the robot bottom operating system,and the specific work is carried out from the following three aspects:(1)Propose a general distributed robot cloud application architectureWe split this cloud applications in several parts,and create a robot remote control and management framework which is a very important part in this architecture.In order to implements this framework we propose a robot remote control model based on virtual agent and a polymorphic mechanism for heterogeneous robot remote control.we realize real-time acquisition and real-time control of robot information by using these model and mechanism.(2)Design a robots application management frameworkBased on the robots remote control and management framework designed in part(1),we design a robot application management framework to realize the overall management of robots.For the implementation of this framework,we design an application description and state transition model,and we also purpose an application management mechanism combined with "dynamic + static".Through these mechanisms and models,we have realized the management and control of static application packages and dynamic applications,and solved the cumbersome management for large-scale robot applications.(3)Design a cloud support platform for distributed robotsWe transform the robots control and management framework designed in part(1)and application management framework designed in part(2)into service,and deploy these services in to the cloud computing platform which was built by Kubernetes in order to provide developers with the management and control of the heterogeneous robots.Finally,a heterogeneous distributed robot cloud support platform is formed,which solves the problem of effective support for large-scale robot management.Finally,based on these architecture and key mechanism,the prototype system is implemented.We have tested the function and performance of the prototype system which verified the feasibility and effectiveness of this work.