Design Challenge 6

Evolve a robot that can draw :
posted by Paul Brown

Summary

This design challenge proposes using evolutionary robotics to make an automaton that can exhibit independent creative behaviour - it can draw (in the non-figurative and fine art sense of the term). Also the resultant drawing style of the robot should not be reminiscent of the drawing style(s) of the people evolving the robot.

The robot is the phenotype and will be controlled by a neural network consisting of a number of inter-linked nodes. The genome consists of a string of floating point numbers where each entry is the weight that attenuates or amplifies the corresponding link between two of the neural network nodes.

A population of genomes is initialised (for example with random values).These are uploaded in turn to the robot that is then tested for its fitness to exhibit the desired behaviour. Fit phenotypes/genomes are allowed to reproduce, unfit ones are culled.

The design challenge is to:

>> identify fitness functions that are requisite to the task of evolving autonomous/independent drawing behaviour.

>> identify  “low level” fitness functions for producing drawings, or their alternatives, or maybe possible routes to their future discovery. See Challenge Details for more on this.