1. Is spare parts recognition via photo already being used today?
Yes, spare parts recognition is no longer just a vision these days, but already a reality. A practical example can be found at Buderus / Bosch Thermotechnik GmbH, manufacturer of heating products and hot water solutions: In the ProScan app, the heating technician selects the device to be repaired and takes a photo of the removed, defective component.
Using technology based on artificial intelligence (AI), ProScan suggests suitable spare parts, which the technician can order directly from the Buderus Online Shop via an integrated interface.
2. How does spare parts recognition via photo and the integration in the digital spare parts catalog work?
We are working on this exciting topic together with Bosch Cognitive Services and offer a full-service model. To enable spare parts recognition via photo, the first step is to automatically take 360° photos of the relevant spare parts. In this process, approx. 300-500 images are created per spare part within 2 minutes.
These are fed into a neuronal database. Bosch's proprietary solution now captures and analyzes the images based on artificial intelligence and algorithms and trains the neural network to enable spare part identification by photo.
As a result, the neural network displays a selection of suitable spare parts with the probability of a match. By integrating it into our solutions for spare parts catalog and service information systems, the service technician can click on one of the results to obtain further information about the corresponding part and its availability and can order it directly online.
Each time the AI solution is used in practice, the artificial intelligence is trained further. It thus becomes more and more intelligent over time, and the quality of the results becomes even better.
3. What are the current limitations and challenges? How can you meet them?
That's exactly what Mathias Bosch, Bosch Cognitive Services, and Daniel Binder, Quanos Service Solutions, talked about in their session at the digital Quanos cDAY 2021.