Menu:
What's New
March 2011:
Get our second take on Slamming Chocolate Bars Episode 2: Kvikk Lunsj, check out on YouTube!
October 2010:
Start of the PAnDA project !
July 2010:
Gene Coded Image (GCI) demo is available.
Research Interests:
Multi-objective Optimisation
As the name suggests, multi-objective optimisation (MO) techniques consider a number of different--potentially even contradictive--goals when optimising a system or process to become better overall. This is an essential technique when tackling real-world and complex problems where it is not possible to rank different goals according to their importance or feasibility.
MO techniques allow to make pareto-optimal choices when selecting solutions that shall be promoted during an optimisation process. As such, MO is a natural way of sampling the search space.
Information on MO can be found here:
Carlos A. Coello Coello
Publications
Stochastic Variability
The building blocks of current FPGAs are strongly affected by intrinsic variability, most severely SRAM and latches. Both process and substrate variations impose major challenges on the reliable fabrication of such devices in deep sub-micron technologies. These variations fall into two categories: deterministic variability, which can be accurately modelled and accounted for using specific design techniques, and stochastic variability, which can only be modelled statistically and is harder to overcome.
Here are some links to related work and my current project:
PAnDA - Programmable Analogue and Digital Array
Device Modelling Group Glasgow
Reconfigurable Architectures
There is a great variety of reconfigurable electronic systems and architectures in both the analogue and the digital domain. The fascinating thing about them is the fact that their functionality is not entirely predefined at the time of their fabrication. Rather, it is possible to develop a range of diffenrent functions in a higher-level design language and map them to such architectures. This makes them very flexible and versatile. For example, it is possible to add new features to hardware systems after building and sending them to customers.
Digital reconfigurable substrates are usually called field-progammmable gate arrays (FPGAs). Analogue reconfigurable substrates are known as either field-programmable transistor arrays (FPTAs) or field-programmable analogue arrays (FPAAs).
Links to reconfigurable architectures can be found here:
FPTA Heidelberg
FPGA Systems
Evolutionary Systems
Evolutionary techniques not only give the potential to explore larger solution spaces, but when implemented on hardware allow system designs to adapt to changes in the environment, including failures in system components. Moreover, bio-inspired optimisation algorithms are global optimisers, which do not require prior knowledge on how to solve a particular problem.More information on evolutionary algorithms can be found here:
Genetic Algorithms Basics
Genetic Programming
Evolvable Hardware
Biological and Artificial Development
I can only agree with this statement. Who would not be amazed by the complex, and sometimes strange, pathways in which tiny little cells develop into large organisms like humans, animals and plants? These mechanisms not even stop working even when the final form has been achieved. They just change their mode of operation and work to constantly maintain the organism, even repair it in the case of damage. How great would it be if we were able to adopt some of these mechanisms into engineered systems?
More information and some examples can be found here:
Developmental Hardware
Intelligent Systems
Genetic Programming (GP)
Cartesian Genetic Programming (CGP)
CGP is a form of genetic programming (GP) that encodes a graph representation of a computer program. It was invented by Julian Miller in 1998. GP, in turn, is concerned with the evolution of computer programs, based on a tree representation of programs. This was inspired by the artificial intelligence computer language, LISP, and the best known figure in this field is John Koza.
More information on GP and CGP is available from this websites:
Genetic Programming
Cartesian Genetic Programming