Cloud native architecture middleware for 5G enhanced autonomous robot

Abstract

This report presents the cloud-native design for a 5G Enhanced Robot Autonomy (5G-ERA) Middleware System. The system is built to integrate vertical applications e.g., autonomous robots handled by the Robot Operating System (ROS), with the 5G infrastructure handled by Open-Source Management and Orchestration (OSM) in the form of Cloud-native Network Applications (NetApps). The cloud-native architecture realised by the Middleware System is an additional layer of the virtual services to facilitate intent-based network and resource orchestration for large-scale deployment of robots under distributed environment. The 5G-ERA Middleware System, which runs on the cloud and edges, allows robots to send data to be processed by distributed robot services, subsequently utilising the computational power of the cloud and edges, and transmitting the outcome of the cloud/edges-based processing back to the robots. This paradigm reduces physical constraints imposed on the robots’ intelligence by local computing resources and simplifies the application development process of distributed robotics. To enable seamless integration, the Middleware System is built with containerised restful APIs, following the microservice and cloud-native paradigms, orchestrated with Kubernetes, and acts as a Kubernetes operator itself by deploying orchestrated applications on demand. The microservices synchronisation problem of the cloud-native design is optimised by introducing the Redis Cluster, a distributed key-value datastore. The contribution integrates local stateless services and a global stateful set for data persistence and paves the way for the Edge switchover required by the connected intelligence.