Intelligent sensor systems

Sensors are key components for the implementation of AI-based applications. In the context of Smart Factory and Smart City we develop intelligent sensor systems. These systems aim to collect process and environmental data in product manufacturing, process engineering, urban areas and for the protection of critical infrastructures, in order to evaluate these data in real time and to provide modern AI applications. As a result, a more efficient and economical operation of production plants and infrastructures can be achieved.

Based on artificial intelligence, we develop intelligent sensor systems for optical quality inspection, which are optimized for easy applicability in medium-sized companies. The result is a very fast and powerful configurable analysis functionality in comparison to classical model-based image processing.

Fields of application

For production plants, we offer solutions such as the smart workpiece carrier INAcarry, which supports to monitor interlinked production plants with only one sensor system. Compared to sensors integrated in stationary systems, the economic advantages of system monitoring are particularly striking. In the domain of Industry 4.0 retrofit, our solutions, such as the INAsense production data acquisition system, facilitate existing machines to be functionally upgraded and integrated into an Industry 4.0 network in order to achieve greater transparency, control, planning and flexibility in production.

Part of the Fraunhofer lighthouse project Machine Learning for Production - ML4P is concerned with the question how extent sensor values can be enriched with process and expert knowledge and how their significance can be increased. The main focus here is to prepare this knowledge in such a way that it can be evaluated by an AI and instructions for optimizing production plants can be derived from it.

Machine learning for industry: At the Fraunhofer Research Center Machine Learning, among other things, intelligent soft sensors are being developed which merge several process variables and generate new, meaningful variables from them. This supports, for example, to detect anomalies in systems which would otherwise have gone unnoticed or to predict the future state of the system.