Optimizing highly constrained truck loadings using a self-adaptive genetic algorithm

Most research into the Container Loading problem has been done on theoretical problem sets and while taking one or two constraints into account. In this paper we discuss the successful implementation of a self-adaptive Genetic Algorithm applying only mutation, with a variable mutation rate. This is applied to a real-world problem with actual problem instances from industry. We introduce an abstract, indirect representation for the considered loadings together with two mutation strategies. Solutions of these different strategies are compared with each other, a static mutation rate GA, and with solutions created by human planners as used in industry, for a set of over 500 realworld problem instances. Furthermore, we examine how our automated results compare to those generated by experienced human planners, showing that they are valid loadings and match fitness values.