Abstract

Genetic algorithms use natural selection and reproduction to search for good solutions to a problem from among a great number of possible solutions. Since they were introduced by John Holland in the early seventies, they have been used in a variety of areas to solve complex problems because genetic algorithms are more robust, more efficient and more flexible than conventional artificial intelligence techniques. Truckin' project uses genetic algorithms to simulate a country where trucks and retailers compete to make profit. The behavior of each truck or retailer is decided by a genome. This thesis does three things on the basis of old Truckin' project. It first enhances trucks' and retailers' performance by adding a new gene to each of them. And then it allows some successful retailers to open new branches to see if other retailers are forced to go bankrupt in the competition. Finally, it analyzes trucks' profit composition and evaluates each gene's performance in the simulation. The simulation result proves the successful use of genetic algorithms in the Truckin' project.