Warehouse Robotics

Order-Picking Optimization as a Service

• Johan Oxenstierna (Lund University's research Portal)

The project concerns the following optimization problems:

1. The Picker Routing Problem (PRP), which is a Traveling Salesman Problem (TSP) adopted within a warehouse environment.
2. The Order Batching Problem (OBP): How to assign orders and compute picking paths for vehicles (sometimes it is also called the Joint Order Batching and Picker Router Problem).
3. The Storage Location Assignment Problem (SLAP): How to assign locations for products.

There is a significant number of publications on these three problems. Within the project, the literature is continuously studied to identify areas where better generalization can be achieved. The first such area that has been identified is that of warehouse layout. The vast majority of proposed optimization algorithms are designed to work for a specific “conventional” layout where racks are laid out in Manhattan style blocks with parallel aisles and cross-aisles. There exists a large proportion of warehouses (still within the "traditional” domain) that do not use this layout, however.

Another identified area is that of computational efficiency. Computational efficiency is relevant because it affects the ease with which an optimization module (e.g. a 3rd party cloud based service) can be integrated with a Warehouse Management System (WMS). This is a topic highly relevant for any warehouse striving to progress into industry 4.0 style operations.

A third identified area is that of benchmarking: There is a scarcity of publicly shared benchmark datasets for the OBP, and an almost total lack of such datasets for the SLAP. There also exists no standard benchmark data format for either problem. Benchmarking is crucial to allow for scientific reproducibility, generalizability, and peer collaboration.