Artificial Intelligence and Data-Driven Methods

Data-driven methods from the fields of Artificial Intelligence, Big Data and Cloud Connectivity are actively used and investigated at IMS in various research projects. Regarding system integration, the use of these methods for selected use cases is being investigated in order to exploit their high potentials in practical implementations at the overall system level.

Modelling is an essential step in the development and optimization of mechatronic systems. Data-driven models can be used to provide useful support when highly complex physical interactions occur, or the exact description of the underlying physical laws becomes inefficient. The change of system properties during operation can also be represented using the continuous recording of operating data and incrementally adapting the models to the data.

In addition, data-based models can be trained with the help of artificial intelligence to support the decision-making process of control tasks, for example to increase overall system efficiencies of powertrains in vehicles, to determine intelligent load distributions of energy systems to simplify the programming of robots and thus making their application more flexible.