Evolvable Hardware Group at FIT, Brno University of Technology, Czech Republic
Summary of Results
In the field of evolvable hardware, evolutionary algorithms (EA) are used to generate configurations for reconfigurable chips that can dynamically alter the functionality and physical connections of their circuits. Research in the field of evolvable hardware can be split into the two related areas of evolutionary hardware design and adaptive hardware. While evolutionary hardware design is the use of EAs for creating innovative (and sometimes patentable) physical designs, the goal of adaptive hardware is to endow physical systems with some adaptive characteristics in order to allow them to operate successfully in a changing environment or under presence of faults.
Here we survey our contributions to the field. Most research papers can be downloaded.
Monographs
PhD thesis
Evolutionary Circuit Design
We have applied evolutionary design to evolve various circuits, in some cases evolved circuits are better than the best known conventional solutions (according to given critera). One utility model was granted in 2009 and a patent application was submitted (see Nonlinear image filter).
Image Filters
Evolved filters exhibit good filtering properties while the implementation cost (e.g. in the FPGA) is very low when compared to conventional solutions such as median-based filters.
Evolutionary Logic Synthesis
Benchmark Circuits for Diagnostics CAD tools
We evolved benchmark circuits with predefined testability and with the size of 1 Mgates.
Image Recognition (with J. Torresen, University of Oslo)
Evolutionary Image Compression (with Ruben Salvador, Universidad Politécnica de Madrid)
Other Circuits
FPGA Accelerators for Evolutionary Design
We have developed the concept of Virtual Reconfigurable Circuits (VRC) for real-world applications of evolvable hardware in FPGAs.
Analysis of the VRC concept
Utilization of Dynamic Partial Reconfiguration (with CEI, Universidad Politécnica de Madrid)
GPU Accelerators for Evolutionary Design
Polymorphic Electronics
We have devoloped first reconfigurable polymorphic ASIC (REPOMO32), implemented several methods for synthesis of polymorphic circuits and proposed several new applications of polymorphic electronics.
Evolutionary Design of Protocols (with P. Svenda and V. Matyas, FI MU Brno)
Evolution using FPTA-2 chip (with NASA JPL)
Development in Evolutionary Design
We have introduced non-trivial mappings between genotypes and phenotypes to overcome scalability problems of evolutionary design. One approach uses an instruction-based development, another one cellular automata.
Cartesian Genetic Programming: theory etc.
Cellular Platforms
Philosophical aspects, theory etc.
Traffic Modelling and Prediction
Survey papers
Last update: Feb 8, 2012