Abstract Faced with high return rates, many e-commerce retailers are considering novel technical solutions to expedite the processing of returned items in their warehouses. One such solution consists of lane-guided transport (LGT) vehicles. These small, electric vehicles follow optical markers on the floor, picking up boxes of returned items at a depot and dropping them off at workstations, releasing the logistics workers to focus on the productive task of actually processing the items instead of carrying them through the warehouse. These types of systems are simple to set up from a technical perspective; however, the routes on the warehouse floor still need to be carefully planned. This gives rise to the following routing problem. Given a set of stations to be served from multiple depots by a fleet of LGT vehicles, which stations doing what type of work should be visited on what route? Only one route per depot is allowed, but multiple vehicles may use the same route. Moreover, since routes cannot be changed on short notice, we consider an infinite planning horizon where the demand rate of the stations depends on the type of work they are assigned to do (e.g., handling defective items or refurbishing). We develop a decomposition heuristic, which solves instances derived from industry data to near-optimality in less than a minute. We also show that the depot location is rather unimportant for the overall system performance, but that the depot count can have a significant influence.

中文翻译：

---

在处理退货的仓库中路由自动车道引导运输车辆

摘要 面对高退货率，许多电子商务零售商正在考虑采用新颖的技术解决方案来加快仓库中退货商品的处理。一种这样的解决方案包括车道引导运输 (LGT) 车辆。这些小型电动汽车跟随地板上的光学标记，在仓库拿起一箱箱退回的物品并将它们放在工作站上，让物流工人专注于实际处理物品的生产任务，而不是将它们运送到仓库. 从技术角度来看，这些类型的系统很容易设置；但是，仓库楼层的路线仍然需要仔细规划。这引起了以下路由问题。给定一组由 LGT 车队从多个仓库提供服务的车站，应该在什么路线上访问哪些车站在做什么类型的工作？每个仓库只允许一条路线，但多辆车辆可以使用相同的路线。此外，由于路线不能在短时间内改变，我们考虑无限规划范围，其中车站的需求率取决于他们被分配的工作类型（例如，处理有缺陷的物品或翻新）。我们开发了一种分解启发式方法，可在不到一分钟的时间内将来自行业数据的实例解析为接近最优。我们还表明，仓库位置对整体系统性能来说并不重要，但仓库数量可能会产生重大影响。我们考虑无限规划范围，其中站点的需求率取决于它们被分配完成的工作类型（例如，处理有缺陷的物品或翻新）。我们开发了一种分解启发式方法，可在不到一分钟的时间内将来自行业数据的实例解析为接近最优。我们还表明，仓库位置对整体系统性能来说并不重要，但仓库数量可能会产生重大影响。我们考虑无限规划范围，其中站点的需求率取决于它们被分配完成的工作类型（例如，处理有缺陷的物品或翻新）。我们开发了一种分解启发式方法，可在不到一分钟的时间内将来自行业数据的实例解析为接近最优。我们还表明，仓库位置对整体系统性能来说并不重要，但仓库数量可能会产生重大影响。