Genetic algorithms for vehicle dispatching

Decisions regarding the dispatch of transportation vehicles for servicing customer requests are mostly based on human expertise. In this paper, genetic algorithms are proposed as a means to support vehicle dispatchers in the context of a courier or express mail service. Vehicle dispatching is first modeled as a multiattribute choice problem. Then, a genetic algorithm is used to find a weighting scheme for the attributes that best reproduces the expert's decision process. Numerical results are reported on data collected from an operations day of a courier service company.