That there is a buzz around artificial intelligence and machine learning is definitely not an overstatement. But is it a temporary hype or a game-changing technology development? And what can it do for warehouse operations? We asked Andreas Anyuru, CTO and Head of R&D at Consafe Logistics to sort things out.
“Today, a lot of people ask themselves: do we have an AI-hype? To start, I’d like to bring up three examples that might answer that question.
Firstly, McKinsey has recently estimated the future economic value created by AI and machine learning to 1.3 trillion USD per year, in supply chain operations alone. That equals Apple’s total market cap, or more than twice the value of Sweden’s GDP. Of course we can’t know whether that estimate is correct, but as we know, McKinsey is a bunch of rather smart people.
Secondly, we can look at the number of participants in the major AI and ML research conferences around the world. For example the important NeurlPS conference has grown from 3,000 to 13,000 attendees in 4 years. This shows a dramatic increase in interest and is also evidence of real scientific progress.
My last example to illustrate the real economic value of AI is a study from Stanford University, comparing the training cost of ImageNet, a certain AI model for image classification, between 2017 and 2018. The result shows an astounding cost decrease, from 2,500 USD to 12 USD in one year. This illustrates the speed of development and AI’s dramatic potential to increase efficiency.
But with all that potential, what can AI be used for when it comes to warehouse management? I’d like to give you a few examples of how AI-based solutions can improve warehouse efficiency.
One is using a so called genetic algorithm, inspired by Charles Darwin’s theory of evolution, to evaluate different packaging solutions against each other and thereby optimize the packaging for each order. This is a way to save freight costs, but is also an important measure to improve sustainability in supply chains by shipping less air. The Swedish postal service PostNord recently stated that according to their calculations standard truck loads today contain up to 30 % air.
Another application for AI-technology is to optimize the placement of goods in the warehouse. By the use of association learning the AI-solution can calculate which articles are often ordered together and put them in the same pick group. The benefits of this are fewer pick rounds, higher efficiency, and shipping fewer boxes.
Another way to increase picking speed is by optimizing the actual route through the warehouse. Finding the optimal route through a warehouse can be seen as an example of the “travelling salesman problem” where an AI-algorithm is used to calculate the shortest possible route for each pick round, instead of using a traditional aisle pick route. This saves time, electricity, and wear and tear on trucks and in general leads to more efficient picking in the warehouse.
And this is just the beginning. In the future we see a number of promising applications of AI in warehouse management, such as automating inbound quality inspection, forecasting product demand, and workforce planning.
To finish off I’d like to return to the initial question: Do we have an AI-hype? Yes, we probably do. But for a good reason. I am convinced AI and Machine Learning will have a dramatic impact on almost all industries in the future. Not least in supply chain operations. Maybe you’ve heard the expression “AI is the new electricity”. What people mean by that is that AI will change our industry in the next 10-20 years in the same way as electricity changed it 100 years ago. Whether this is true we don’t know yet, but it sure looks like it.”