Bicycle-sharing systems expansion

Abstract

Bicycle-sharing systems (BSSs) which provide short-term shared bike usage services for the public are becoming very popular in many large cities. The accelerating bike traveling demands from the public have driven several significant expansions of many BSSs to place additional bikes and stations in their extended service regions. Meanwhile, to capture individuals' traveling needs more precisely, in the expansion, many BSSs have set up online websites to receive station location suggestions from the public. In this paper, we will study the bike station re-deployment problem in the BSSs expansion. Besides the historical bike usage and construction cost information, the crowd suggestions are also incorporated in the problem. The station re-deployment problem is very challenging to solve, and it covers two sub-tasks simultaneously: (1) bike station locations identification, and (2) bike dock assignment (to the deployed stations). To address the problem, a novel bike station re-deployment framework, CrowdPlanning, is introduced in this paper. In both station deployment and capacity assignment tasks, CrowdPlanning fuses different categories of spatial information including the crowd suggestions, individuals' historical bike usage and the construction costs simultaneously. By formulating these two tasks as two optimization problems, the optimal expansion strategies can be identified by CrowdPlanning. for the BSSs. Extensive experiments are conducted on the real-world BSSs and crowd suggestion dataset to demonstrate the effectiveness of framework CrowdPlanning.