Unsupervised Clustering Method for the Capacited Vehicle Routing Problem

In this paper an unsupervised clustering method for the Capacited Vehicle Routing Problem is proposed. Advantages and disadvantages of the proposed algorithm are weighed, and comparisons are made to some clustering algorithms commonly used in the literature to tackle routing problems. Experimental tests were performed using Solomon and Hering/Homberger benchmarks applied to different distributions. The proposed algorithm is demonstrated as effective for the attempted problem, with the ability to improve upon weaknesses in some of the clustering algorithms in the literature.