Aalto researchers find a new method for task allocation in distribution centers

FIMECC/EFFIMA project “Control of information in Multi-Machine environment” (MOTTI), has studied a generic multi-machine system from a system point of view. One of the cases in MOTTI has been order picking in distribution centers. Order picking is a multi-machine application, which makes possible to machines execute tasks simultaneously.

A new method for task allocation saves 18 000 km in order picking warehouse.

In order-picking the pickers drive around the warehouse and pick items according to customer orders. The machines are driven by humans, however humans are basically computer controlled through voice commanding system. It is a very labor intensive operation and can cause as much as 50% of the total warehousing operating costs. Furthermore, most of the order processing time is spent only on driving around the warehouse.

“Basically, each picker has three baskets to which the items are picked. You can not touch the content of a single basket, because it belongs to one customer order. However you can select which baskets you are giving to one picker”, says Jari Saarinen, the project manager of MOTTI, who works at Aalto University in the Generic Intelligent Machines Centre of Excellence (GIM).

This gave the motivation for MOTTI researcher Marek Matusiak, also working at Aalto and in GIM, to investigate possibilities of re-organizing the task execution system so that the traveled distance is optimized. During a short time period there are several tens of orders that are processed. Combining orders that has items in a same area will logically reduce the need for traveling between picking.

“This is a combinatorial problem that is extremely heavy operation even for todays computers. Finally, we were able to find a way to reduce the combinatorial space, which gave us a way to estimate higher number of batches”, Marek Matusiak says.

The new method saves 18 000 km driving

Researchers received a three months dataset from an order picking warehouse and they recomputed all the routes for the pickers using the method, which in average resulted 15% of savings.

“The other way to put the savings is that annually the savings is about 18 000 km, which means that for the given warehouse you would have to hire two tow-truck drivers just to drive around the warehouse to fulfill the saved kilometers”, says Matusiak. Also the partner in cooperation is satisfied with the result.

“This is an ideal method for our Visual Assistant product, which we have partly developed during the project”, comments Team Manager Jani Mähönen from Rocla. With Rocla’s Visual Assistant picker receives information about several picking locations in line, so he can drive in the middle of them and collect all the items with only one stop. Using this visual display in picking work reduces unnecessary stops up to 80 %. “The method developed in MOTTI groups the collected items even closer together, which will even further increase the performance of the Visual Assistant. We are really looking forward to test this in practice”, continues Jani Mähönen.