inicioir al DIIbuscar
courier delivery in the presence of uncertainty
Vehicle routing in many industrial applications is faced with significant uncertainty which can make a carefully planned routing solution inefficient. In this work we consider the problem of routing courier delivery vehicles to satisfy a random customer demand, random service times and time windows. In addition to standard objectives, such as travel times and time window violations, courier services desire to operate on stable routes that do not change much from day to day, due to training and reliability concerns. For this problem, we propose a hybrid model combining robust optimization and scenario-based stochastic programming to obtain a stable master route and daily scheduling solutions. We present results that show that robust vehicle routing models can protect against the uncertainty at a modest additional cost and thus provide stable routes. We also develop a two-phase insertion-based heuristic to solve industry sized problems. Our experiments show that our heuristic improves similarity of routes and total penalty at the expense of increasing route lengths compared to independent deterministic solutions for each demand outcome. We also show that our solution can outperform the current industry practice on a real-life problem.