The SERVICEFACTORY project, a joint effort of adidas AG, Deutsche Telekom AG, Dresden Elektronik Ingenieurtechnik GmbH, Humotion GmbH, RWTH Aachen, DFKI, and VDI, aimed to create a digital service ecosystem for sports and fitness. The initiative focused on integrating heterogeneous sensor data, developing a cloud‑based analytics platform, and delivering personalized recommendation services.
Technical Results
Central to the project was the design of a connectivity architecture that links body‑area networks, such as the adidas Smart Shoe and Humotion performance‑diagnosis belt, to a cloud platform hosted on Microsoft Azure. Data streams from these devices are enriched with contextual metadata (e.g., activity type, location, and user profile) before being anonymised and made available through a unified data service. This abstraction layer enables downstream analytics without imposing rigid data schemas.
The analytics deployment on Azure incorporates a recommendation engine built on tensor factorisation techniques. The engine processes aggregated sensor, context, and user data to generate product‑to‑product, product‑to‑user, and user‑to‑user recommendations. Figure 15 of the project documentation reports the achieved performance of this approach, demonstrating significant improvements over baseline collaborative filtering methods.
Several smart services were instantiated from the recommendation engine. The Shoe Recommendation service (illustrated in Figure 12) suggests optimal footwear based on gait analysis and training load. The Playlist Recommender (Figure 14) delivers tempo‑matched music tracks to runners, with a prototype app shown in Figure 16. A Performance Analysis service (Figure 17) aggregates physiological metrics to provide real‑time feedback on training effectiveness.
Hardware prototypes such as the Smart Shoe (Figure 18) and Smart Jacket (Figure 19) were integrated into the platform, enabling continuous monitoring of biomechanical and environmental parameters. The IPSO smart object framework (Figure 20) further standardised data capture across devices. An API for subsession data extraction (Figure 21) facilitated third‑party application development.
Overall, the project delivered a modular, cloud‑based service factory that can ingest diverse sensor streams, apply advanced analytics, and expose actionable insights through a suite of recommendation services. The architecture supports scalability and real‑time processing, positioning it for broader adoption in the sports‑tech market.
Collaboration and Impact
The consortium combined expertise from industry, academia, and research institutions. adidas AG supplied real‑world sports equipment and user data, while Humotion GmbH contributed performance‑diagnosis technology. Deutsche Telekom AG provided connectivity solutions and cloud infrastructure, and DFKI led the data mining and recommendation engine development. RWTH Aachen and VDI contributed methodological support and quality assurance.
Work packages were clearly delineated: TP1 focused on data modelling and service development, TP2 on sensor data aggregation and analytics, and TP3 on hardware integration and platform interfaces. Regular roadshows, such as the adidas event (Figure 23) and the Smart Shoe station showcase (Figure 24), demonstrated the platform’s capabilities to stakeholders and potential customers.
The project’s outcomes are positioned for commercial exploitation through the Servicefactory ecosystem (Figure 8). By bridging the gap between industrial digitalisation and consumer‑facing services, SERVICEFACTORY exemplifies a holistic approach to smart‑service development that can be replicated across sectors.