Optimizing crowdsourced delivery routes through concurrent selection of pickup stores and drivers

Many retailers are offering crowdsourced delivery where ad hoc drivers collect goods from pickup stores and deliver the goods to customers on behalf of the retailers. Efficient spatial data management solutions are needed to optimize the routes of the drivers. In the existing model of crowdsourced delivery, retailers make delivery plans regardless of drivers' original full routes. This can lead to unnecessarily high delivery costs. We propose a novel crowdsourced delivery model that optimizes delivery routes by concurrently selecting pickup stores and drivers. Based on this model, we develop a heuristic solution for crowdsourced delivery. Experimental results show that our solution saves delivery costs by up to 50% compared with the existing model used by retailers. Our solution is also scalable for large cities.