The Bus Touring Problem: Cluster - first, route - second heuristic

The Bus Touring Problem (BTP) objective is to maximize the total attractiveness of the tour by selecting a subset of sites to be visited and scenic routes to be travelled - both having associated non-negative attractiveness values - given the geographical frame considerations, constraints on touring time, cost and/or total distance. Real-life large-scale BTPs can be modelled, but practically they are not solvable, being NP-complete. Thus effective heuristic methods have been developed, aimed at generating a fast near-optimal BTP solution. Unfortunately, there are specific cases in which these existing heuristics may either consume a considerable amount of time or produce poor results when solving large-sized touring problems. This paper presents a radically new approach to cope practically with real-life large-scale Bus Touring Problems: a new heuristic based on clustering first, then routing. The results of a real-life touring problem are presented.