Delivering AI to the manufacturing industry.
We now turn our attention to how KITT4SME intends to deliver AI to the manufacturing industry through its purposely designed workflow. The KITT4SME workflow comprises five high-level activities: Diagnose, Compose, Sense, Intervene and Evolve. The figure below shows a simplified version of this workflow which is adequate for our purpose of introducing the subject. However, the reader is encouraged to examine the details of the workflow in the original proposal.
The first step in the workflow is the Diagnose step where given the current state of the shop floor, the system determines, through a smart questionnaire, how AI can help to transition the shop floor to a more desirable state. Following that, in the Compose step, out of a graph of software components, their wiring and configuration, that is the marketplace catalogue, a minimal sub-graph is selected which is suitable to achieve the desired transition to a better state. Then this tailor-made software kit is deployed to the cloud as a set of services.
As soon as the shop floor is connected to those services, data start flowing from the factory to the cloud where the tailor-made kit detects and explains anomalies and visualises key performance indicators (KPIs). These activities comprise the Sensing step. Perhaps, it might help to think of the shop floor as a large dynamical system where one would like to measure its current state with some degree of accuracy in order to use the dynamics function (think AI model) to predict if the shop floor is en route for a desirable or undesirable outcome, as shown in the picture above with a sad emoticon face.
In the Intervene step, the system suggests and negotiates corrective actions to steer the shop floor towards the desired production and staff well-being goals. Finally, in the Evolve step, the system analyses the outcome to improve the Diagnose and Compose steps, and to propose personalised staff training.