Using metaheuristics for the location of bicycle stations

TL;DR

The paper addresses optimal bicycle-station placement by framing it as a $p$-median problem and evaluating five metaheuristics (GA, ILS, PSO, SA, VNS) with automatic parameter tuning via irace on realistic Malaga data. GA consistently delivers the best solutions, with real-distance metrics and population-based demand weights producing substantial reductions in walking distance compared to the current system. The study demonstrates that smartly adding stations can yield large usability gains (roughly $33$–$53\%$ reduction in travel to the nearest station) with significant implications for smart-city planning. The approach provides a practical, data-driven framework for urban planners to improve shared-bike infrastructure and can be extended to related facility-location problems.

Abstract

In this work, we solve the problem of finding the best locations to place stations for depositing/collecting shared bicycles. To do this, we model the problem as the p-median problem, that is a major existing localization problem in optimization. The p-median problem seeks to place a set of facilities (bicycle stations) in a way that minimizes the distance between a set of clients (citizens) and their closest facility (bike station). We have used a genetic algorithm, iterated local search, particle swarm optimization, simulated annealing, and variable neighbourhood search, to find the best locations for the bicycle stations and study their comparative advantages. We use irace to parameterize each algorithm automatically, to contribute with a methodology to fine-tune algorithms automatically. We have also studied different real data (distance and weights) from diverse open data sources from a real city, Malaga (Spain), hopefully leading to a final smart city application. We have compared our results with the implemented solution in Malaga. Finally, we have analyzed how we can use our proposal to improve the existing system in the city by adding more stations.

Figures (6)

• Figure 1: The centre of each neighbourhood (orange points) and current bicycle stations in the city of Malaga (blue points).
• Figure 3: Fitness value obtained in each of the scenarios for each algorithm.
• Figure 4: Empirical cumulative distribution of the percentage of improvement of our solutions in each algorithm and scenario, compared to the current real location of bicycle stations in Malaga. We evaluate for each scenario the current solution of Malaga using the specific type of distance and weight in each case.
• Figure 5: Average distance (m) walked by the citizens to reach its nearest station. Each algorithm is shown in each combination of distances and weighting model.
• Figure 6: Smooth function over the average distance walked by a citizen to its nearest station in the solution obtained in each iteration of the four betters algorithms. We decided to exclude PSO in order to clarify the figure, as its lowest value among all scenarios was 1,018.24 m.