Knowledge4Retail

Supporting retail with digital twins.

With its platform, Knowledge4Retail (K4R) is advancing the development and use of AI and the use of service robots in retail. Here, so-called "semantic digital twins" (semdZ) of shops serve as the basis for all applications. With this technology, retailers are given the opportunity to align their assortment even better with the wishes of their customers and to make the connection between online and offline shopping more effective. For example, services can analyse customer behaviour in the shops and optimise the placement of goods in terms of visibility and accessibility based on this. Other services will enable automated inventory and shelf replenishment with the help of service robots. This will strengthen the stationary retail trade in the long term.

Market perspective and product promise

The emerging Knowledge4Retail platform is intended to drive the dissemination and development of AI-supported hardware and software services in the retail sector with the help of standardised data formats, interfaces and solutions. The aim is to operate the platform in the long term via a coordinating organisation, for example in the form of a company supported by the project partners. The platform is aimed at all submarkets of the retail industry, such as drugstores, supermarkets, clothing shops or DIY stores.

Consortium

team neusta GmbH, German Research Center for Artificial Intelligence GmbH, dm-drogerie markt GmbH + Co. KG, dmTECH GmbH, EHI Retail Institute GmbH, fortiss GmbH, Fraunhofer-Gesellschaft zur Förderung der angewandten Forschung eingetragener Verein, Nagarro AES GmbH, neusta GmbH, neusta software development West GmbH, Ubimax GmbH, University of Bremen, Technische Universität München.

Challenge and innovation

In retail, most strategic decisions are made based on customer buying behaviour. By capturing incoming and outgoing goods, the product range and prices can be adjusted to increase sales figures and turnover. What has been captured very little so far, but has high potential, is the analysis of the concrete behaviour of customers in the shops. AI can shed light on the "black box" of retail outlets. For this purpose, the K4R project will bring together targeted data on a platform for the first time.

Solution approach

The platform is based on semantic digital twins - digital models of real shops that combine different data of a retail shop. For this purpose, the project is developing a data format for the digital representation of the structure and processes of retail shops. What is innovative here is that not only is corresponding data collected for the models, but this data is also linked semantically with each other. The digital twins are based, for example, on data from shop sensors or ERP systems, which are enriched with other information systems such as digital product catalogues. The data is also supplemented by as much data as possible from the environment of the branches, such as geodata, to enable a wide range of AI applications. In the project, first pilot applications will be designed and provided by developers.

Through standardisation, an interaction of data suppliers and solution providers is possible. Based on this emerging ecosystem, developers can further improve their AI applications or combine them with applications from other providers to create new AI applications.

In addition, retailers can use their semantic digital twins to easily integrate autonomous service robots. These then navigate independently in the shops based on the digital models. At the same time, the robots feed new sensor information back to the platform.

Use Cases

Intelligent intralogistics

Based on the semantic digital twin of retail shops, the intralogistical process from the warehouse to the shelf is optimised with AI applications. The semdZ can reflect the exact inventory and location for the placement of newly delivered goods. AI helps to intelligently arrange the goods for each sales floor section so that stocking becomes more efficient. It also identifies empty shelves, which are then prioritised for stocking. The semantic digital twins also help to optimise offers such as "Click & Collect", i.e. the collection of customer orders in the shop.

Strategic retail marketing to build customised shops

In order to keep pace with online retail, stationary retail must tailor its products more to local customer needs. Newly developed AI algorithms suggest an optimised composition of the store assortment. Through continuous updates of the semdZ of each shop, the results of the AI methods are constantly improved.

Service robotics to support store employees

In order to relieve employees in intralogistics, an autonomous robot is being developed that can be loaded in the warehouse and independently travels to the target shelf to be unloaded there by the employee. For its navigation, it uses the data of the digital twins.

Internet of Things (IoT) connection of a smart refrigerator.

The transferability of the K4R platform to other industries is demonstrated using the example of a smart refrigerator in merchandise management. The smart fridge interacts with the K4R platform and can be opened cashlessly via app, customer card, credit card or employee card. It is able to independently recognise, track, manage and sell its inventory and thus demonstrates the future scenario of fully automated sales concepts.