Package examples.dlife.robot.player.evolution

An example of how Player/Stage simulations can be use to evolve robot controllers.

See:
Description

Class Summary

EvolvedObstacleAvoidance	Illustration of how to evolve a neural controller for a robot.
HemissonNeuroController	Controller that controls the Hemisson using a provided neural network.
HemissonProxSensorFilter	Filter the Hemisson's proximity sensor values so that they make nice inputs to a neural network.

Package examples.dlife.robot.player.evolution Description

An example of how Player/Stage simulations can be use to evolve robot controllers. The classes in this package illustrate how the dlife.robot.player, dlife.ga and dlife.nn packages can be used together to evolve a robot controller to perform obstacle avoidance. Please note that this example is provided as an illustration, not a practical example. As such, it contains a very small population (2 controllers) that is evolved for a very short time (5 generations) with very short lifetimes (5 seconds). Thus, it is unlikely that actual obstacle avoidance will evolve unless these values are increased.

See the EvolvedObstacleAvoidance class for instructions on running the example in this package.