Hyper-Heuristics for Irregular Object Multi-Container Packing

This paper presents a preliminary study on the use of hyper-heuristics for solving multi-container irregular object packing problems. The work is part of a larger project which aims to develop a robotic packing system for the application of packing nuclear waste into containers. As part of this project, a novel packing optimisation scheme is proposed to plan how to pack the objects into given containers. We propose a new hyper-heuristic algorithm for optimising both packing order and placement heuristics for each object. We analyse the performance and highlight the strengths and weaknesses of the proposed approach. The work presented in this paper is conducted on 2D datasets, however all the methodology can be adapted to 3D packing. Based on the comparison of our results against a 2D multi-container packing algorithm from literature, our approach shows promise, however it suffers primarily from needing large computation times to find good solutions owing to the large number of combinations to the problem. Future work will focus on ways to cut down the number of combinations (without degrading the quality of solutions) as well as trying to speed up the packing algorithm by limiting the number of rotation angles for the objects.