Reliable indoor positioning (<10cm) enables several innovative location-based services, because such accuracy levels essentially allow for real-time interaction between humans and cyber-physical systems. Activity recognition, machine navigation (e.g., “shelf” level), geo-fencing, and automated robotics; are among services that yield safety-critical assembly processes and logistics. Specifically, for a production process that demands the involvement of humans and robots to assemble heavy and complex entities like car engines or power supply units, with robot assisting on carrying these heavy products for assembly, to prevent collisions and accidents, a real-time indoor localization service that monitors the flow of objects and detect the collaborating human workers positions is mandatory for enabling a safe working environment.

Indoor positioning for safety-critical industrial IoT requires the propagation of telemetry, positioning and trajectory data at millisecond range from hundreds of thousands of objects, human workers and robotic machinery. Effectively monitoring these entities requires numerous sensors, including: wireless/lidar/ultrasonic distance measuring modules, temperature/humidity/pressure modules, cameras, and accelerometers. These sensors can be mobile or mounted (e.g., factory floor/wall/ceiling). At the same time, it requires the execution of complex algorithmic models on 3D-spacing topologies to output coordination plans, continuously assess and prevent collisions among objects, robotic machinery and workers for specific factory sections and assembly lines. Because of the delay-sensitive nature of these tasks, propagating acquired positioning data to centrally accessible private cloud infrastructure, results in cycles, where often due to either unanticipated load and model processing, the outputted safety distances, coordination assessment and planning are derived too late. Too late means a cycle output results with a response rate >100ms. To understand the importance of <100ms cycles, consider a worker mistakenly swaying his/her hand in the potential operating area of a moving machine. This results in assembly lines, or even entire factory sections, to immediately halt to prevent collision and tragic events with numerous halts affecting overall production, assembly line resetting (due to quality and security re-evaluation), machinery quality (due to non graceful halts) and human worker psychology.

As the time sensitive nature of the real time indoor positioning system, the collection and processing of the sensor data cannot be decoupled. For a small application area, one powerful central machine might be enough to process all collected data, however this still comes with the challenge of creating fast enough communication channels between this central server and the server nodes. This affects the flexibility and scalability of the application greatly. Furthermore, the central server might be overloaded in high activity times where there is a high number of workers in the manufacturing floor, while staying idle but continue consuming power during low activity times. To mitigate this, Fog/Edge computing for each application area is utilized to decentralize the collection and processing of the data to each area and scale up and down depending on the current workload, e.g amount of machinery or number of workers in the area.

RAINBOW provides the necessary orchestration capabilities for creating, managing and scaling these Fog/Edge application areas. Its dashboard helps visualizing the current resources while helping to extend the resources by adding new devices to the environment. Furthermore, RAINBOW enables processing and visualization of the telemetry and performance statistics. These statistics then help detect the critical areas of the manufacturing floor, therefore enabling targeted upgrades when needed. By orchestrating redundancy with the use of RAINBOW, any potentially harmful major failures are prevented by quickly replacing failed nodes or when there are no more enough resources, graceful shutdown of the system with insightful analytics.

0 Comments

Leave a reply

Your email address will not be published. Required fields are marked *

*

Rainbow Project ©2022 All rights reserved

Log in with your credentials

Forgot your details?