Robust Workforce Management with Crowdsourced Delivery

We investigate how crowdsourced delivery platforms with both contracted and ad-hoc couriers can effectively manage their workforce to meet delivery demands amidst uncertainties. Our objective is to minimize the hiring costs of contracted couriers and the crowdsourcing costs of ad-hoc couriers while considering the uncertain availability and behavior of the latter. Due to the complication of calibrating these uncertainties through data-driven approaches, we instead introduce a basic reduced information model to estimate the upper bound of the crowdsourcing cost and a generalized reduced information model to obtain a tighter bound. Subsequently, we formulate a robust satisficing model associated with the generalized reduced information model and show that a binary search algorithm can tackle the model exactly by solving a modest number of convex optimization problems. Our numerical tests using Solomon’s data sets show that reduced information models provide decent approximations for practical delivery scenarios. Simulation tests further demonstrate that the robust satisficing model has better out-of-sample performance than the empirical optimization model that minimizes the total cost under historical scenarios.