Conventional IT systems are rule based and correspond with static messages. In other words, we define functions and decision trees that machines then implement. However, today’s businesses want more. They need highly flexible and tailored IT support for a variety of order types and structures received from across multiple sales channels. For example, systems must simultaneously support deliveries to brick-and-mortar shops and e-commerce fulfillment. A first step is to deploy algorithms that optimize fulfillment. SSI SCHAEFER has deployed these for the client Desigual, allowing us to continuously adjust and improve order management processes and to improve the cost efficiency of picking. Although, the way logic is currently modeled in IT systems limits what we can do currently.
At SSI SCHAEFER, our innovative culture keeps us forward-thinking, and AI is our next big mission. We want to create intralogistics processes that are able to adapt to customers’ changing needs – autonomously and dynamically. We want IT systems that are “open-minded” and able to identify and analyze patterns; for instance in ordering behavior, in situations where humans, with our focus on causal links, cannot process. This paradigm shift will allow us to design processes that are more agile, and more situation-sensitive. This will enable us, for example, to predict customer orders before those orders are placed, and to perform picking and commence shipment at an earlier stage. We want to harness the customer knowledge that resides in our data.
AI is the attempt to give computer systems the ability to think for themselves to a degree. However, even the experts cannot really agree upon an exact definition of intelligence. One example is neural networks. This is an IT model of cognitive structures with the aim of approximating a generally unknown functional correlation between input data and outcomes. These systems analyze possible connections and use the data available to them in way that is not possible by humans with preconceptions and prejudices. Or, put another way, people think in terms of specific problems. Machines look for connections, and give us answers to questions that we may never have posed in the first place.
AI systems that have been trained to perform corresponding intralogistics tasks can help human workers in warehouses. AI systems will make recommendations and improve process efficiency by using smart forecasts. Prescriptive maintenance, for instance, will allow the early forecasting of the remaining service life of a given machine. Potential faults will be diagnosed in advance and preventive maintenance performed with the support of proactive intralogistics processes. In other words, this will minimize machine downtime. Prescriptive maintenance combines the intelligence of both hardware and software.
In the 1990s, we simply did not have the volume of data or the processing power for sophisticated machine-learning processes demanded by our imperatives. Today’s hardware and high-performance chips make it possible. Big data technology puts us in the comfortable position of being able to supply systems with artificial knowledge and enable continuous learning. Deep learning, i.e. a type of machine learning based on hierarchical neural networks, is now proven and viable. Ultimately, we have more flexible, improved abilities. We can go beyond theory, and deploy these technologies in our day-to-day work. Things are continuously evolving and getting better, greatly expanding our AI possibilities.
We have seen our IT and software skills develop at SSI SCHAEFER IT Solutions to open up new AI opportunities. SSI SCHAEFER will soon be able to deploy AI systems in customer projects – at least as it pertains to historical data. In the future, this will lead to us doing less programming work, and concentrating more on training systems, with the goal of greater project success. At the same time, we need to retain control over the system. How far do we go? We will require fallback strategies that allow us to respond to unforeseen changes on the customer side. This is the only way of ensuring that the customer’s warehouse is always operational. Ultimately, we want to implement a solution where the only limitation is the physical intralogistics equipment itself.
Flexibility is the central characteristic of software support, allowing greater responsiveness to customer needs. However, to achieve a successful AI project, communication between data science, simulation professionals, and the people who are actually responsible for implementation is key. Understanding the real-world customer situation is paramount as SSI SCHAEFER leaps forward.