Introduction
 

The design productivity gap in the electronic industry is a well known fact. How can the design community utilise the design capacity that technology is offering and at the same time ensure it's correctness? In the search for solutions for a given task or problem, the problem has to be specified. The solution to the problem is often dependent on the specification of the problem as this steers the implementation of the solution. This places a requirement of accuracy on the specification. If the problem is very complex or incomplete then the required level of accuracy of the specification can be hard to achieve. To find solutions to the problem of developing large and complex designs new design paradigms are required \cite{road97}.

One possible solution is to turn away from traditional design techniques following the various design and testing phases and instead allow hardware to evolve until a correct solution is found. This technique is termed hardware evolution or equally, evolvable hardware. Artificial evolution is a technique that may be applied to problems where the problem space is too large for exhaustive searching. Evolvable hardware may be considered to be a subset of artificial evolution, where the evolved solution is represented in hardware instead of software.

Two main methods have been established for applying artificial evolution to the design of hardware systems. These are Extrinsic and Intrinsic evolution.. In Extrinsic (off-line) evolution, the evolution process and the resulting evaluations are implemented in software. Each individual, design instance generated from the evolution process, is evaluated by a software simulation of the design described by the individual. When evolution is complete, the resulting design needs to be implemented in hardware.

Intrinsic (on-line) evolution, takes the design process closer to the real hardware in that each individual is tested out in hardware i.e. not an abstraction of it. Although the evolution process is still implemented in software, assessment of design quality is based on an actual implementation. When evolution is complete, the resulting design is already implemented in hardware.

In our work, a third method termed 'Complete Hardware Evolution' (CHE) is also used. Instead of having all (Extrinsic) or part (Intrinsic) of the evolution process in the host processor, a hardware implementation of the evolution process is used to drive evolution.

[main page]

Faculty
    Snorre Aunet, Assistant Professor, Department of Computer Science and Informatics
    Pauline C Haddow, Associate Professor, Department of Computer Science and Informatics

PhD Students
   Frode Eskelund, IDI Financing
    Morten Hartmann, IDI financing
    Katherina Jørgensen, NT financing
    Per Kristian Lehre, IDI financing
    Gunnar Tufte, NFR financing
    Piet Van Remortel, (joint with VUB) Belgian financing

Current masters students
    Lars Thomas Boye
    Jan Sigurd Dragsjø

Faculty Partners
    Berit Johansen , Asssociate Professor, Department of Botany, NTNU
    Astrid Lægreid, Associate Professor, Department of Physiology and Biomedicine Technology,  NTNU

Associated groups
     Jim Tørresen, Evolvable Hardware, University of Oslo
     Jan Kommorowski

[main page]

Adaptive and Fault Tolerant Hardware
    Digital Filtering
    Robotics
    Messy Gates

Future Technologies

Artificial Development: Shrinking the Genotype

Hardware Modelling of Biological Processes

Complete Hardware Evolution

[main page]

Student Projects

Masters Projects
   Applying Evolution to incomplete information in a biological setting, Jan Sigurd Dragsjø
   Using Artificial Development to aid evolution of modelled biological processes, Lars Thomas Boye

Completed Masters Projects

    Evolution of Robust Circuits in a Simulated Enivronment, Frode Eskelund
    Evolutionary Fault Repair of Electronics in Space Applications, Sverre Vigander [supervised of Adrian Thompson]
    Prototyping of Complete Hardware Evolution, Gunnar Tufte
    Configuration of a Virtex FPGA for Evolvable Hardware,  Espen Tislevoll
    Routing in a Mulitprocessor using Evolvable Hardware, Knut Helge Vindheim
    Adaptive Hardware based on Evolution, Morten Skoglund
    Design of RF/IR Interface between a Robot and External Equipment, Tor Arne Olaussen
    Variation of Selection for GA, Mathis Landsverk

Completed Final Year Projects

   Evolution of Fault Tolerant Digital Systems, Andreas Engh-Halstvedt og Frode Eskelund
    Representasjon av biologisk enheter, Lars Thomas Boye og Jan Sigurd Drasjl
    EHW in the Microarray Project, Dag Kristian Rognlien
    Co-evolution, Cat and Mouse project, Mathis Landsverk and Geir Martin Hynne
    GERC, a Genetically Evolved Robot Controller, Espen Tislevoll and Morten Hartmann
    Adaptive Mutation: Controlling a Parameter in Genetic Algorithms, Sverre Vigander

[main page]