Concept and Methodology
Concept
5G-ERA is conceptually concerned with experimentation facilities able to provide enhanced experimentation infrastructures on top of which, third party experimenters e.g. SMEs or any service provider and target vertical users will have the opportunity to test their applications in an integrated, open, cooperative and fully featured network platform running across multiple domains where needed, and tailored to specific use cases related to robotic applications. To explore this concept, the project will use existing 5G testbeds in experimentation facilities which will be used to verify the robotic applications through a number of use cases.
To achieve autonomy, a robot must ‘learn’ during its operational process. Because of the limited computing resources, safety constrains and biased training data, the learning needs will be shifted to the cloud using 5G technologies. The 5G-ERA workflow follows PDCA (plan–do–check–act) model where “plan”, “do” and “check” are carried out virtually in clouds and the “act” is performed in a real workspace by the robot. By trialling virtually, it avoids safety constrains. Additionally, the virtual environments are always updated with data coming from the real workspace enabling the so-called “Digital Twin” (of the Robot, of workspace and operators) that can be used not only for online real-time warnings and feedbacks, but also to carry out a what-if analysis as well as support the decision process and/or the optimization of the robot’s operational environment. Thanks to 5G technologies, cloud and robots can be linked closely together for collective intelligence where the robot sends its sensing data to the cloud and updates the local decision model via the cloud.
Methodology
- 01st
Challenge
Optimising the QoE of 5G orchestrators for vertical applications
- 02nd
Challenge
Optimising the testbeds towards Cloud Native (CN) approach for scalability, availability and feature velocity required by the use cases
- 03rd
Challenge
Extending 5G open environment and standard APIs of testbeds into robotic vertical sectors
Further development and standardization of experimentation facilities are both highly important and urgently needed in the development of autonomous robotic applications across vertical sectors. 5G Enhanced Robot Autonomy (5G-ERA) addresses the experimentation facilities via the following activities:
- integrating operational processes of essential autonomous capabilities into Open Source MANO (OSM), ensuring the vertical specific adaptation of existing experimentation facilities
- realisation of an intent-based networking paradigm by aligning the end-to-end (E2E) resource optimisation with the autonomous operations, ensuring effective policy to be designed
- Cloud native Network Services (NSs) on the experimental facilities will create, ensuring robotic applications exploiting NFV/SDN infrastructures efficiently
- extending the experimentation facilities into robotic domains thorough standard APIs under Robot Operating System (ROS), prompting third-parties’ experimentation activities as well as engagement from new players
The methodology also encompasses achieving multi-domain capabilities verification through a specific workflow procedure involving multiple domains from multiple providers. Each domain has its on way in terms of exposing resource virtualisation and automated provisioning of resources to services. For example, as illustrated in the figure below:
- Domain 1 is the private cloud with a few customised microservices of robotic capabilities mainly under ROS
- Domain 2 is MEC which holding pre-existing knowledge for local workspace, it is shared by robots visiting the workspace. MEC is also managed by chains of OSM
- Domains 3 and 4 are commercial cloud infrastructure providing additional resources required by large scale machine learning
- Domain 5 is the external data centre for knowledge discovery and replay, they are not necessary to be maintained by the same parties
- Robots are controlled using ROS
