Abstract: This paper solved the three-dimensional rectangular packing problem with a quasi-human approach.By defining the maximal rectangular spaces at current iteration, the action space, we improved our caving degree approach such that the computation is largely speeded up at the same time the excellent characteristic of the caving degree is still kept.In this way a good solution could be achieved in a shorter time. In the experiments, we tested the improved algorithm with 47 without-orientation-constraint instances in the OR-Library. Computational results show an average space utilination of 95.24%, which improves current best result reported in the literature by 0.32 %. In addition, the results also show less running time compared with other algorithms.

Key words: NP-hard,Threcdimensional packing, Heuristic, Quasi human, Caving degree