AI algorithms and methods unsuitable for embedded systems
Usually, embedded systems are specifically adapted to their task such as data acquisition or control or signal processing in a device and optimised for minimal cost, space, energy and memory consumption. The numerous existing AI algorithms and methods, in contrast, usually operate on much more powerful, larger, more energy-intensive and more expensive computing technology and are unsuitable for embedded systems. In addition, there are often too few data sets available for applications to train an AI, such as from fault situations on a machine for which predictive maintenance is to be supported with AI.
We optimise AI algorithms and methods for embedded systems
We are researching to optimise AI algorithms and methods so that they can be used on embedded systems. For predictive maintenance applications, for example, we are working on generating artificial data, increasing robustness and transferring the solutions to similar problems.
Dr.-Ing. Tino Hutschenreuther
Head of System Design
tino.hutschenreuther(at)imms.de+49 (0) 3677 874 93 40
Dr. Tino Hutschenreuther will answer your questions on our research in Smart distributed measurement and test systems and the related core topics Analysis of distributed IoT systems, Embedded AI and Real-time data processing and communications, on the lead applications Adaptive edge AI systems for industrial application and IoT systems for cooperative environmental monitoring as well as on the range of services for the development of embedded systems.
In thurAI, IMMS is working on sensor technology for SmartCity and methods to intelligently process data in the network for AI evaluations.
IMMS has developed an AI-based, retrofittable and cost-effective solution for quality assurance of machining tools.
To use compressed air for industrial processes in an energy-efficient way, IMMS has developed the electronics platform for an automatable sensor solution.