Based on current architecture mentioned in the blog post “RAINBOW Human-Robot Collaboration in Industrial Ecosystems Use Case” there are certain requirement/criteria that become crucial when the architecture needs to scale and cater to larger needs. The following are the high-level requirements that need to be satisfied by RAINBOW:
- Scalability of cloud-native services and effective utilization of resource on Fog device. In a typical industrial environment, the number of personnel and robots in a work-place area keeps changing. As PLMC, RMT, CPA service only supports fixed size group of personnel and robots, there is a need to dynamically scale instances of this service on-need basis, typically based on movement of personnel or/and change in workplace configuration of robots.
- Deterministic and bounded system latency. To predict and avoid collision between personnel and Robot, the PLMC, RMT, CPA services must be able to perform their respective processing tasks and send appropriate signals to Robots to stop in case of possible collision prediction in a deterministic timeframe with a hard upper bound. The term of this this upper bound for timeframe is System Latency. In an event of predicted collision, this system latency plays a key role in determining the “safe distance” (a minimum distance between personnel and robot to avoid collision). In the scenario when system latency is violated, it can cause unscheduled interruption in production (by stopping of robots) or in worst case can cause fatal collision between personnel and robots. Thus, there is a need to monitor the Service Level Objectives (SOL) continuously. Taking corrective or preventive measures based on policies to ensure SOLs are met with utmost importance.
- Dynamic resource provisioning at runtime. In the scenario when a fog device in an infrastructure fails to serve many instances of PLM, RMT, CPA services due to lack of resources (typically refers to computational, storage, network resources), there is a need to dynamically provision these resources from near-by Fog/Cloud devices (if and only if prescribed SOLs can be meet).
- Reliable dynamic service provisioning between Fog devices. In a typical industrial environment, personnel move from one work-place area to other. The personnel’s position is received by Aggregator device from mobile Node device mounted on the personnel through wireless communication. A major pitfall for wireless communication physical layer is that it has limited range of coverage. When a person moves from one workplace coverage area to other, the Aggregator device responsible for receiving these telemetry messages changes and exactly one Aggregator is associated to one Fog device running Collision Prediction and Avoidance services. Since the position of the personnel changes within different workplaces, the processing Fog device also changes. Thus, there is a need to reliably provision a new service instance running in Fog device within the new workplace area and then transfer the data stored in the local database of old workplace to new workplace Fog device. Once data is transferred successfully the service instance running in the Fog device of old workplace is terminated.