Vehicle routing problems arise in many practical situations in the context of transportation logistics. Among them, we can highlight the problem of transporting customers from origin to destination locations, which is known as the dial-a-ride problem (DARP). This problem consists of designing least-cost routes to serve pickup-and-delivery requests, while meeting capacity, time window, maximum route duration, and maximum ride time constraints. This work proposes a hybrid algorithm to solve single and multi-depot DARP variants where both the demands and vehicle fleet are heterogeneous. The method combines the iterated local search metaheuristic with an exact procedure based on a set partitioning approach. In addition, several procedures were implemented to speedup the local search phase. Extensive computational experiments were conducted on existing and newly proposed benchmark instances in order to evaluate the impact of the different components of the algorithm, and to compare its performance against the best existing method. The results obtained suggest that the proposed algorithm is capable of producing highly competitive results regarding both solution quality and CPU time, and of improving several best known solutions.

在运输物流的背景下，在许多实际情况下都会出现车辆路径问题。其中，我们重点介绍了将客户从始发地运送到目的地的问题，即所谓的“搭便车问题”（DARP）。这个问题包括设计成本最低的路线来满足取送请求，同时满足容量，时间窗口，最大路线持续时间和最大行驶时间限制。这项工作提出了一种混合算法，用于解决需求和车辆车队都是异构的单点和多点DARP变体。该方法将迭代的局部搜索元启发式方法与基于集合划分方法的精确过程结合在一起。此外，还实施了一些程序来加快本地搜索阶段。在现有和新提出的基准实例上进行了广泛的计算实验，以评估算法不同组件的影响，并将其性能与现有的最佳方法进行比较。获得的结果表明，所提出的算法能够在解决方案质量和CPU时间方面产生极具竞争力的结果，并且能够改进几种最著名的解决方案。