The fourth industrial revolution is fueling unprecedented cyber-physical convergence and will ultimately redefine entire industries and ruthlessly undermine companies that fail to transform. Data from physical systems, such as machines, meters and sensors play a critical role in enabling cyber-physical convergence to thrive. However, this data can be notoriously challenging to harvest and analyze, because of its volume, sparsity, and multi-dimensionality. This is often exacerbated by the real-time nature of many high value digital services and the need to aggregate data and intelligence amongst machines and devices.
To effectively harvest, analyze and act upon data and accelerate cyber-physical convergence, sophisticated Edge AI capabilities are required, which include:
· Embedded AI, capable of on-device inference and learning, to efficiently harvest, analyze and act upon voluminous, sparse, multi-dimensional and real-time data from individual machines and devices, which have constrained compute footprints.
· Standardized APIs so that the data and intelligence can be aggregated across multiple embedded AI installations, and;
· Integration across operational technology (OT) and amongst OT and IT layers to effectively create Edge AI hierarchies. With these hierarchies, the harvested data can be acted upon for different purposes depending on the OT and IT demands.
The stakes are high for embedded system providers, OEMs, software and systems integrators and end users, as the fourth industrial revolution takes hold. Embedded and Edge AI can no longer be an after-thought and must be incorporated in future system designs from the outset. Complacency is not an option, and with the right design, stakeholders across entire value chains can leverage embedded and Edge AI for competitive advantage.
Dr Phil Marshall, Chief Research Officer Topio Networks
For more on the topics, download our white paper