Dr. Dirk MayerFraunhofer IIS/EAS
Dr. Dirk Mayer
Design and Validation Support for Trustworthy IIoT Sensors
In industrial applications, sensors are a key element to enable added value for data analytics and process automation, paving the way to real autonomous, adaptive and smart systems. Modern sensor systems are compact, energy-efficient and less expensive, while providing condensed information with ultra-low latency. Integration of such sensors in industrial environments is an enabler for novel IIoT applications. Examples are the integration of condition monitoring systems into structural elements of machinery or autonomous systems in robotics and intralogistics. However, particularly industrial scenarios like online process optimization or dynamic closed-loop control for robotics impose high demands for trustworthiness and reliability on the applied sensors systems. Without a trustworthy design and verification concept, this would not be feasible, neither technically nor economically. This poses new challenges when developing and integrating sensors. Validation procedures for these complex micro-mechatronic systems have to be in place. In addition, training and testing of integrated data pre-processing and machine learning algorithms is required.
Trustable design and verification methodologies for such IIoT sensor systems is one of the main innovation fields researched by Fraunhofer IIS/EAS via its Application and Test Center for AI (ATKI). The goal of ATKI in Dresden is offering consulting services and design and verification support for AI-based systems such as IIoT sensor system.
In this presentation, two demonstration scenarios for sensor applications are introduced, that require trustworthy output data:
Integration and test of MEMS sensors into industrial condition monitoring applications and virtual development of perception sensors for autonomous mobile robots.
Concepts for virtual and experimental validation methods such as Sensor-in-the-Loop are discussed, that reduce the amount of costly field tests using the real machines within the real environment. This enables rapid design and validation spins when exploring new trustable sensor applications.