The Industrial Metaverse will enable operators of industrial facilities to monitor activity and optimize performance. This in turn will enable increased automation at a time when labor costs are growing rapidly. Multiple components are needed to ensure that a digital representation of an industrial facility can add value.
- There must be a network of multimodal sensors to measure the structure of the facility and detect changes within it.
- We require a physical world digital twin to represent all the activity currently taking place within the facility.
- We require a business-process digital twin to connect the physical-space representation with the business processes related to the facility, e.g., supply-chain, inventory management, demand forecasting etc
- In many cases there will be an ecosystem digital twin that spans multiple commercial entities and connects their separate in-house digital twins.
In the AI Research Lab at Nokia Bell Labs, we are focusing on real-time physical world digital twins that are kept up to date using drone-based industrial monitoring. The construction of this dynamic digital twins has two phases. In the first we use a scan from a lidar-equipped phone or standalone camera to create a high-fidelity point cloud of the basic structure of the facility. In the second we fly a drone in the facility and apply analytics to detect and localize objects in the images streamed from the drone. The digital twin listens to the signal transmitted by the drone and places a digital copy of the object in the twin, whenever a corresponding object is detected in the real facility. In the other direction, changes made by an operator in the digital twin can be communicated to robotic agents who then carry out these changes within the facility.
A dynamic digital twin (left) representing boxes recognized by a drone flight (right)
One application of this system is warehouse monitoring. The drone counts boxes on the warehouse shelves and reads the barcodes so that the boxes can be labeled within the digital twin. The system connects to the warehouse monitoring system and allows the warehouse operator to track goods as they enter and leave the warehouse, and locate any unused space which enables more efficient storage.
In another application our system monitors plants in an urban farm. The drone detects areas of poor growth, measures the height of the plants, and detects critical events in the grow-cycle such as germination and the time that the leaves of the plants start to overlap.
