Experimentation System for Determining Bus Routes for Customers of Supermarket Chains

It may be observed that supermarket chain companies invest in special buses to deliver customers directly to a given store. The companies expect a tool which allows to design the routes of buses in order to minimize costs (maximize profits). The objective of this paper is to present the created computer experimentation system (simulator) with the designed and implemented algorithms to determine the optimal bus route for customers. The bus route problem was divided into two stages. At the first stage, the three designed heuristic algorithms called Most Occupied, High Gain Neighbor, and Cut the Worst are responsible for selection (choosing) of the location of bus stops. At the second stage, the five algorithms allow to determine a route between previously chosen stops. These algorithms are based on approaches from so-called Artificial Intelligence area: Simulated Annealing (SA), Taboo Search (TS), Genetic Search (GS), and Ant Colony (AC) as well as on the simple ideas like Random Search (RS). In the paper, the results of the investigations made with the created experimentation system are presented, concerning an adjustment of the parameters of any algorithm, and a comparison of the algorithm’s efficiency on both stages.