Collaborative Crowdsourced Vehicles for Last-Mile Delivery Application Using Hedonic Cooperative Games

Abstract

In this paper, the problem of collaboration in crowdsourced last-mile delivery is addressed, where multiple crowdsourced vehicles cooperate to fulfill tasks. Collaborative crowdsourced frameworks allow recruited vehicles, referred to as workers, to perform shorter trips while expanding the geographic coverage. Existing solutions in collaborative, crowdsourced last-mile delivery solely maximize the task allocation without considering 1) cost factors such as travel distance and payoff and 2) the self-interest of crowdsourced workers. As a solution, we propose a hedonic cooperative game approach that determines delivery routes and assigns relaying vehicles by maximizing the average payoff per kilometer, where payoffs are based on task contributions. Specifically, the proposed algorithm, hedonic crowd relay assignment (HCRA), uses the Nash equilibria of a series of hedonic games as the basis for the task allocation. To compute the workers’ preference lists, HCRA relies on crowd relay breadth-first search (CR-BFS) to find a set of potential routes for task completion, given the constraints of the vehicles. The proposed solution is compared to a benchmark, and the results demonstrate that a more efficient and scalable solution is achieved using HCRA, where both the workers’ total payoffs and average payoff per kilometer are increased, even with increasing numbers of vehicles, tasks, and relays.

Keywords

• Collaborative algorithm
• cooperative hedonic game
• crowdsourced vehicles
• last-mile delivery
• task allocation