Crowdsourced task dispatching for the shared electric vehicle relocation problem: a hybrid variable neighbourhood search and genetic algorithm

Shared electric vehicle relocation (SEVR) is essential to the shared mobility and cost-benefit of a one-way, floating-station vehicle-sharing system. This study investigates the crowdsourced task dispatching problem for SEVR to rebalance the spatial variation in supply and demand under random demand. The objective is to minimize total costs by dynamic matching among relocation tasks, stations, and crowdsourced dispatchers. Single-task crowdsourcing (STC) and multitask crowdsourcing (MTC) dispatching models are presented. A hybrid algorithm, combining an improved genetic algorithm with variable neighbourhood search, is devised to solve the problem. Scenario analysis, using trajectory data of electric taxis in Changchun City, shows that compared with the benchmark algorithm, the hybrid algorithm improves the solution quality by 5.09% and reduces the running time by 31.20%. STC is preferred over MTC in the low supply-to-demand density (R) scenario, though MTC is not rejected. However, MTC performs more effectively in the medium and high R scenarios.