Utvikling av altruisme med gjenkjennelse av slektskap i evolusjonære roboter

Swarm robotics provides ways of solving problems using groups of autonomous robots that display emergent intelligent behaviour. Control mechanisms for these robots can be designed by using mechanisms from evolutionary robotics. In order to ensure that the controllers are adaptive and capable of learning new behaviour controllers are artificially evolved using techniques inspired by natural evolution. Altruism can be helpful to ensure that the controllers solve problems in a way that maximizes the utility of the population as a whole. Kin-selection and kinrecognition have been shown to be beneficial for evolving altruism in nature. This report investigates if kin-selection and kin-recognition can be used to ensure the evolution of altruism in evolutionary robotics. An experiment is conducted by giving swarm robots controllers evolved with a genetic algorithm. The robots are given the ability to relinquish fitness and different mechanisms for discerning kin. The results give support to the hypothesis that kin recognition can be used to evolve altruism in evolutionary robotics