Unknown Artificial Intelligence for Lifting and Learning

It is against this backdrop that EGOKIA has emerged. The project aims to develop a type of AI that does not depend on the stability of the environment, but is instead equipped to cope with change: an AI that detects when a process is changing, is capable of characterising that change, and learns or adapts without halting production or requiring comprehensive retraining. To this end, EGOKIA addresses the complete detection–characterisation–adaptation cycle within the Open World Learning (OWL [2,3,4]) paradigm: identifying that something new is happening, understanding whether that change is relevant to the industrial task, and adapting in a controlled manner to maintain process reliability. All of this is achieved by building mechanisms and functionalities that can be adopted in real industrial environments, with hardware constraints and stability requirements.