
Photo: Palto – shutterstock.com
The images of interrupted supply chains are present in the public consciousness like never before: empty shelves in supermarkets, warehouses from which medical material disappears on pallets overnight, truck traffic jams at closed borders and entire production facilities that stand still because individual parts are missing. The challenges for those responsible:
-
Monitor logistics processes seamlessly,
-
Identify weaknesses at an early stage
-
and provide alternatives at short notice.
After all, it is a matter of restarting the attacked economy as quickly as possible and making it more resistant to disruptive influences – both in small and large. Robotic Process Automation (RPA) is a quickly implementable solution for interface problems at the many transfer points in every supply chain. In addition, methods and tools based on Artificial Intelligence (AI) not only support better demand forecasts, but also data-driven plan changes in real time.
To be able to solve a problem, you have to know its cause. This apparently trivial finding often becomes a serious challenge when it comes to logistics. Because many of those involved do not know who is involved in which role in the supply chain – the crisis has also shown that. The transfer of data between the systems that control the process is often monotonous, error-prone manual work. Due to the sheer mass of the shipments, important controls are only carried out on a random basis. As a result, errors in the supply chain often only become apparent when the recipient’s production has stopped.
The example of a large European logistics company in the field of air freight shows how digital media can do better. Here, software robots check all flights on which the company’s shipments are booked every few minutes. This enables digital employees to identify deviations at an early stage, for example when a particular aircraft changes course. Depending on the cause and duration of the deviation, they also propose suitable measures to get the consignments concerned to their destination as quickly as possible. In any case, the responsible dispatcher receives all the necessary data in real time to make quick decisions and inform the customers concerned. This means that they can use their technicians elsewhere in the meantime, for example if an urgent spare parts delivery for technical customer service is delayed.
This real application scenario is made possible by a supervised machine learning algorithm based on a decision tree. Without a bot, the task for the dispatcher would not be manageable due to the amount of data that must be continuously processed within a very short time. The status messages of the flight monitoring to be checked alone amount to more than 2000 per minute at normal times. In addition, the most suitable one must be selected from thousands of possible combinations of shipments, flight movements and freight capacities in order to then initiate the appropriate measures.
For bots, it is no problem to collect and process almost any amount of data from a variety of systems according to a predefined process – error-free and around the clock, seven days a week. If you combine advanced algorithms from predictive analysis with operations research methods such as simulations or game theory models, digital logistics employees are able to plan with foresight. But even without AI involvement, RPA delivers tangible added value for logistics companies: At DHL, for example, software robots send transaction-related emails, bill carriers, plan appointments to ensure timely deliveries, and call up delivery confirmations.
Capacity planning is a key success factor in logistics. The more precisely those involved forecast the demand early and look at it under various restrictions, the more reliably, robustly and cost-effectively they can provide the required capacities on time. The problem: Although there is a considerable excess of transport capacity worldwide, there is often no suitable cargo hold where it is needed. In extreme situations such as the recent crisis, this leads to the collapse of the supply chains. The suspension of passenger flights during the lockdown caused a dramatic shortage of air freight capacities.
Usually, more than 50 percent of air freight – including urgent medication deliveries – is carried on passenger flights. Thanks to the bots used, it was possible to find the few flights still taking place in a very short time and to get at least some of the most urgent deliveries on the way. Apart from this extreme case, there are a multitude of different influencing factors and restrictions to be considered in the capacity planning of logistics: In addition to the general economic situation, these include:
-
Consumer mood,
-
Fashion trends,
-
government regulations,
-
Weather conditions,
-
Fuel prices
-
or also political developments.
In order to make this complexity manageable, advanced algorithms therefore become an elementary part of logistics planning: They evaluate historical data from several years from numerous different information systems and combine them with transaction data from company applications and (bought or freely available) information from market research and social media.
In a German freight forwarder, for example, they provide demand forecasts for the next week – with an accuracy of more than 70 percent. This saves the company cash every day because every empty trip costs. And in view of an empty journey share of over 37 percent and almost 6.6 billion empty kilometers in German freight transport (in 2018), it is clear that there is still enormous potential here by optimizing the loading space using algorithms. Intelligent systems for price formation on logistics platforms such as Timocom, Transporeon or Saloodo make an important contribution because they bring supply and demand together more quickly.
Important when implementing such projects: In order to train the respective algorithm adequately, the historical data of at least three years is usually required – in first-class quality. The systems also have to cope with explosive growth in the movement data. A single autonomous automobile generates up to 30 gigabytes of data within one hour of operation. Around 500 million posts appear on social media – on Twitter alone. In addition to RPA and AI, data management will play a crucial role in tomorrow’s logistics.
- Steve Oluborode, Tableau
“The most important steps are to standardize the structure of self-made databases and data formats, to dissolve redundancies and to bring them into a format. In addition, there are skills and resources as well as a modern data culture. Companies cannot achieve their goals with an isolated initiative. They need to close ranks IT, the departments and the management level – everyone has to work hand in hand so that the departments can quickly access the relevant data. “ - Petra Pirron, Datavard
“Many companies need a translator between business and IT in order to achie
ve a consensus on requirements and services and to bridge the gap between backend and visualization. One goal is to make previous investments in SAP landscapes, analytics tools, data lakes or Protect the cloud and bring it together with a sense of proportion – what do I have today, what do I need today, and what will I need tomorrow? The following applies: There is no ‘one size fits all’ in data management. “
Last mile logistics shows how inevitable digitalization has become for the functioning of logistics processes: Here, algorithms decide on location planning and loading of depots. Because no matter whether Next Day or Same Day Delivery: For providers, these concepts are only economical if they can predict exactly at any time when which goods in which depot and in what quantity must be available. This is especially necessary when stationary retailing is only possible to a limited extent.
The operational procurement process in the supply chain is characterized by routine activities that bots can perform quickly and without errors. In warehouse management, for example, they already trigger requirements based on situation and rules based on inventory information and generate orders independently. In supplier communication, they transfer data between different systems without implementing a complex transmission process such as EDI. This enables companies to quickly integrate new partners into their processes and prevent disruptions in the supply chain.
In approval processes, including checking incoming invoices or quality certificates, software robots compare actual data with target data – for example from supply contracts – and enable their human colleagues to concentrate on cases with deviations. The prerequisite for these and other automation steps is the consolidation and continuous maintenance of master data in the different company applications, which can also be made more efficient with the help of bots.
In view of the multitude of areas of application for bots and AI, the question remains of how the cooperation between people and digital employees can be meaningfully coordinated. An important aspect against the background of the shortage of skilled workers: Well-trained junior staff who do not want to spend their time doing routine work can prove their knowledge and skills in a variety of demanding tasks. This motivates and strengthens their loyalty to the company.
The challenge for managers in the future will be to promote and optimally use employee skills such as solution orientation, empathy, communication skills, analytical and conceptual thinking or judgment. This also includes the targeted use of digital employees who work with human colleagues. Because people make the essential decisions in tomorrow’s logistics. (mb / fm)