Insect Division of Labour Applied to Online Scheduling

Scheduling in the painting facilities of automotive manufacturing plants is a complex process. The paint loss caused by machines continuously switching colours leads to significant costs. Optimising these schedules while assigning new jobs in real time and dealing with machine break downs is therefore an important issue in the truck painting problem. Previous successes in solving this problem with algorithms inspired by colony insects made evident that this is an area worthy of further investigation. Both improving on existing methods and looking from other angles through previously unconsidered models of division of labour inspired by insects may lead to solutions closer to the the optimum. Improving the understanding of the strengths and weaknesses of various approaches to scheduling for painting trucks is done by comparing them on a singular problem set. Along with the previously used methods and yet unconsidered models is a proposed algorithm building on those existing approaches. To measure the performance of all the used approaches the total setup time, throughput and flow time were kept track of. At the same time queue and storage usage were measured to avoid solutions with unrealistic requirements. As newly considered model, foraging for work showed a phenomenal performance in terms of the total setup time. This was primarily the case in experiments with uniform colour distributions and those with heavy workloads, which the model was able to handle best across all measured statistics. For a realistic colour distribution it was also competitive on the front of total setup time. While the proposed algorithm could not compete with the performance by foraging for work on the total setup time, it showed improvement over the methods it used as inspiration in most situations. With a realistic colour distribution the proposed algorithm was also competitive with foraging for work on total setup time and outperformed it on most other measures. Foraging for work is presented as a new and important competitor in optimising for the total setup time. Further, the importance of evaluating the performance of new approaches which may excel in different areas deserves emphasis. The proposed algorithm serves as another step forward in decreasing setup times while maintaining a high throughput.