The main objective of the project is to save all the current barriers in the design and implementation of Evolvable and Adaptive Hardware (EAH) and make it an indispensable element in all those future electronic devices that are subject to degradation processes, either by their own functionality or as a result of the interaction with other degradable devices. Medical measuring devices or drug delivery systems, which physically interact with the human body, are a clear example of this type of systems. All our research in EAHs will be validated through the implementation of a customizable and Evolvable Artificial Pancreas (EvoAP) system. The major milestones to be accomplished are:
(1) The implementation of new EAH devices, wearable and autonomous. These devices will integrate modules designed using Darwinian principles. They will be designed using Computational Intelligence techniques either in the design flow or adapting their behaviour in real-time.
(2) To attain reachable and scalable circuits. To this end, we will start from our recently published proposals on adaptive and evolvable filters and the implementation of evolvable controllers based on model induction.
Additionally, the real-world validation of the two previous points, will lead us to two additional milestones:
(3) The induction of new models of blood glucose levels in humans using techniques from Genetic Programming field. This will provide a new framework, unexplored nowadays, to define non-linear models that include particular features of each patient, leading to personalized models.
(4) The development of a complete EvoAP system. It will integrate all the devices conceived in (1) and (2), as well as the models defined in (3). Current state of the art artificial pancreas fail mainly in the impossibility to recover from sensor degradation. On the contrary, EvoAP will provide the adaptability as well as fault-tolerant features, needed on faults in the hardware or in the sensors and electronics of the devices.