Aarhus Universitets segl

Distributed Computational Intelligence

Computational systems become increasingly pervasive in our environment. Whether this is in our private sector, such as a smart homes, wearables, and the Internet-of-Things, in the public sector, e.g. smart transportation and autonomous cars, or in the industrial sector, for example smart production systems, industry 4.0, and smart farming.

These systems, independent from their application, are usually designed to operate in very specific situations and under pre-defined conditions. While they are often able to adapt within certain boundaries, they struggle in completely new, unexpected, and often quickly unfolding situations and do not perform to their full potential. With the rising number of devices deployed in our environment, a central control or common coordination approaches for all devices will not be feasible. We therefore require systems able to learn individually about their situation, their environment, and themselves and use this knowledge when interacting and self-organising with others in order to optimize their individual but also their collective performance during runtime. This brings together computational intelligence and distributed systems in distributed computational intelligence.

In this research group, we study distributed computational intelligence, where individual autonomous, intelligent systems interact in order to achieve common goals. We are tackling problems with approaches inspired by biological processes observed in nature in combination with ideas from autonomous and self-aware computing systems. While this covers a wide area of research, the current focus is on autonomous and self-aware systems able to learn about themselves and their environment in order to cooperate, collaborate, and potentially self-organize with other systems to overcome new, unexpected, and rapidly unfolding situations.

Our vision for the future is having systems, deployed without any a priori knowledge, able to learn about their environment autonomously. Establish collaborations with other systems during runtime without common controllers and pre-defined coordination mechanisms. Eventually, these systems will be able to explore new situations and define their own goals and problems to be solved making maintenance and deployment efforts obsolete.