A tree search heuristic for the container retrieval problem

Abstract

The problem of finding a sequence of crane moves in order to retrieve all containers from a block of container stacks in a predefined order with a minimum of crane move time is called the container retrieval problem. In this article, we describe a tree search heuristic based on a natural classification scheme of crane moves that is able to solve the container retrieval problem for typical problem dimensions in very short computation times. A comparison with the approach recently published by Lee and Lee (2010) shows that the heuristic is a very competitive method for solving the container retrieval problem. The mean crane move time for the tested problem instances could be reduced by 15.7 %, the average number of crane moves was reduced by 8.7 %, while the computation speed could be drastically improved from several hours to less than 10 seconds on average, the latter being especially important for the application in container terminals where the retrieval decisions have to be made in short periods of time.