Evolvable hardware

The project concerns the use of evolvable hardware for real-word and real-time applications. Since a couple of years, new technologies have been developed that make reconfigurable hardware devices possible, i.e. devices of which the archetecture and as a result its functionality can be changed extremely fast while the device is still functionning.Since their reintroduction in the mid-eighties adaptation or learning techniques such as neural networks, genetic algorithms became very important and have been applied succesfuly in many areas such as computer science, engineering, economics, artificial life. One major drawback of these techniques is that the adaptations have to proceed offline insteadof online because the underlying adaptation algorithms are currently too slow. Oggline means that the adaptive system is trained in a test environment which is a simplification of the real environment and it can use as much time as needed to adapt to that environment. Later on, the system is then put in the real environment. A much better and realistic situation is an adaptive system which adapts in real-time to the real environment, i.e. online adaptation. The combination of reconfigurable hardware devices with the existing adaptation techniques offers new perspectives for online adaptation and in this project we want to explore this combination further.