Adaptation in natural and artificial systems
Adaptation in natural and artificial systems
Co-evolving task-dependent visual morphologies in predator-prey experiments
GECCO'03 Proceedings of the 2003 international conference on Genetic and evolutionary computation: PartI
| Hi-index | 0.00 |
My research has been in the domain of mobile robots with Learning Classifier System (LCS) controllers. LCSs[1] utilize reinforcement learning and a genetic algorithm to evolve a set of condition-action rules, or classifiers. If a particular condition matches the state of the environment, then the action is executed. In a robot system, sensors monitor the environment, while actions commonly correspond to wheel velocities. Rule sets are initialized with random elements, and, in a process mimicking natural selection, the best sets are carried forward to the next generation. Genetic operators, such as crossover and mutation, create changes within the rule sets, creating a dynamic learning system. To test our LCS implementation we conducted several experiments in the pursuit domain.