SNCF, the French public train company, is experimenting to develop new types
of transportation services by tackling vehicle routing problems. While many
deep learning models have been used to tackle efficiently vehicle routing
problems, it is difficult to take into account time related constraints. In
this paper, we solve the Capacitated Vehicle Routing Problem with Time Windows
(CVRPTW) and the Capacitated Pickup and Delivery Problem with Time Windows
(CPDPTW) with a constructive iterative Deep Learning algorithm. We use an
Attention Encoder-Decoder structure and design a novel insertion heuristic for
the feasibility check of the CPDPTW. Our models yields results that are better
than best known learning solutions on the CVRPTW. We show the feasibility of
deep learning techniques for solving the CPDPTW but witness the limitations of
our iterative approach in terms of computational complexity.

使用注意力编码解码结构和一种新颖的插入启发式算法，我们以构造迭代深度学习算法解决带时间窗口的车辆路径问题，并比最佳已知学习解决方案更好地解决了该问题。