A Simulated Annealing Algorithm to Solve the Multi-objective Bike Routing Problem

With the increasing concern about sustainable means of transport, the use of bicycle is earning popularity. However, for cyclists, it may be a challenge to use this mean of transportation in an urban context, since there are plenty of route choices available and each of them may represent a different trade-off between multiple objectives, such as safety and travel distance. These trade-offs depend on cyclist's experience and preferences. Thus, while an experienced cyclist may choose the shortest path between the source and destination points, a beginner may prefer to travel longer distances to cycle in dedicated cycling infrastructures or to avoid motorized traffic. In this paper, it is proposed a simulated annealing algorithm, to achieve an approximated Pareto set. One uses the A-Star algorithm to create an initial solution and propose a perturbation method, which uses the A-Star algorithm, to generate a higher diversity of solutions. The methodology was tested in practical use cases and was able to find a set of quality solutions for the bike routing problem (BRP).