Evolutionary Approach for the Multi-objective Bike Routing Problem

In this paper, a multi-objective approach for the bike routing problem is presented. Bike routing represents specific challenges, since cyclists have different experiences, concerns, and route preferences. Our approach considers two criteria: the total traveled distance and the cyclists safety. Finding the optimal Pareto set is computationally unfeasible for these problems, therefore, the goal of this work is to create a non-exact method capable of producing a set of quality solutions in a timely manner. A heuristic that modifies the multi-label setting algorithm is used to create an initial population and a genetic elitist algorithm is used to find an approximated Pareto set of optimal routes. The proposed methodology is applied on a practical case study, in which real data from OpenStreetMaps (OSM) and Shuttle Radar Topography Mission (SRTM) was used to model the graph for the road network of the city of Aveiro, with 9506 nodes and 21208 edges. The results show that the approach is fast enough for interactive use in a planning tool and produces a set of quality solutions, regarding two criteria, the traveled distance and the safety of the path.