Bosch Research and the AIT Austrian Institute of Technology (competence unit Complex Dynamical Systems / Center for Vision, Automation & Control) are strengthening their long-standing successful collaboration in the field of "Federated Learning" over the next three years.
Already today, algorithms for the automation of mobile machines are being trained by machine learning. The training of the necessary models can take place decentrally on the control units of the individual machines (local learning). The use of Federated Learning will also make it possible in the future to combine models and thus knowledge about the behaviour of several machines in the cloud into an improved overall model. In this way, individual machines that have collected data under extreme environmental conditions such as sub-zero temperatures can pass on the "learned" knowledge to other machines. Another example of cross-machine knowledge transfer is the machines' ability to detect malfunctions in their operating behaviour or defective components because they deviate from "normal behaviour".
Read more:
Contact
DI Amadeus-Cosimo Lobe
Complex Dynamical Systems
AIT Center for Vision, Automation & Control
+43 664 88256044