Artificial Evolution(AE), inspired from the famous theory of evolutions of Darwin, is a relatively new approach for solving complex problems in fields ranging from engineering and robotics to social sciences and genetics.
AE approach relies on computing the "fitness" (a.k.a. quality ) of a population of candidate solutions for a given problem. Based on the fitness values of the candidates, a new generation of candidate solutions are reproduced using genetic operators like recombination and mutation. The candidate solutions are then used to generate "offsprings" based on their fitness values; fitter candidates generating more offsprings into the next generation.
Artificial evolution has also been used in robotics, both for evolving controllers for autonomous robots as well as for evolving their morphologies. However, one limiting factor in these applications is the high computational cost of "fitness computation" of solution candidates. In robotics, these evaluations are typically done by a realistic simulation of the system, and is costly in terms of computation. Hence, the fitness evaluations of candidate solutions need to be spread over a group of computers for speeding up the process.
This work aims to create a Grid-based framework for AE applications. The framework should create a transparent interface for the AE user which would distribute the execution of the evolution onto the Grid.
During the analysis, design and implementation of the framework, we will use evolutionary robotics problems (such as developing controllers for autonomous robots) as our main application domain for guidance.