Is automated item picking right for me?

THESE RETAIL warehouses are also more likely to have complex order profiles, requiring multiple SKUs to fill an order. As a result, they are under the most pressure to adopt new processes and technologies that enable faster order cycle times, reduce costs and alleviate challenges created by tight labour markets.

Pick-and-place robots designed for material handling are similar in some basic ways to the robots that have become common in manufacturing environments. However, the demands on robotic systems in material handling are very different than those in manufacturing. In manufacturing applications, robots are performing repetitive activities using highly standardised parts or components. In item picking, robots must recognise and handle a broader range of products sizes, shapes and weights.

To meet the demands of material handling applications, pick-and-place robots require a higher level of sophistication in some key systems than robots used in manufacturing.

Providing automated picking support for goods-to-person automation systems is a very promising application for pick- and-place robots that could significantly improve warehouse operations. To be effective in this application, the robot must be able to effectively pick a significant percentage of products stored in the automation system. This is possible in a growing number of applications due to the advances in robotic vision and gripping systems that occurred in the last several years.

The current generation of robotic item pickers have demonstrated the ability to recognise and handle a wide range of products without special testing. Products must, of course, be small enough to be graspable by the gripper. In the case of the Swisslog AutoPiQ solution, the gripper can handle products with dimensions of 25 mm by 20 mm with heights up to 300 mm and weights to 1.5 kg. Other characteristics that facilitate effective gripping include stable and closed objects/packaging, carton or hard plastic packaging with regular surfaces, and packaging with surfaces not permeable to air.

In addition, machine learning enables robots performing item picking to learn from experience. Using adaptive algorithms, the robot can improve its picking performance based on the data it collects and using that data to detect patterns, make associations and gain insights. The more “experience” the robot gains in an application, the more data it has to learn from and the better it can perform. This experience can even be shared across sites and companies to accelerate the learning process. Machine learning is being integrated into the current generation of robotic item pickers, which will enable these systems to pick faster and more efficiently the longer they are in operation.

Working with AutoStore

AutoStore represents an ideal platform for integration of pick-and-place robots. The system is highly standardised, which enables specific application considerations to be precisely replicated in the lab. This allows throughput to be accurately projected and a business case for robotic item picking to be developed. 

In most cases, the business case will be driven by cost savings resulting from increased productivity of human pickers or a reduction in the number of man hours required to complete a particular task.

Generally, the automation solution must be designed to the technology. Item picking robots aren't as flexible as humans and solution design should account for this. In addition, the WMS must support the robots through capabilities such as order splitting, volume calculations and expanded error handling. Tight integration of all components in the solution is crucial to overall performance.

To download the white paper go to https://bit.ly/3chhSpm