Decision-making support for suitable AI hardware
The question for other companies is which hardware to use for the data-intensive processing of algorithms if calculations are not to be carried out in the cloud – for example, in the event of strict data-protection requirements or a need for real-time responses. So that businesses can make well-founded decisions on their specific use of AI, the AI Application and Test Center offers tailor-made consulting and services with a focus on three key aspects:
1. hardware performance for the processing of AI tasks
2. edge AI integration
3. data reduction when using edge computing
Comprehensive and manufacturer-independent comparisons of current AI hardware, as well as practical tests, allow quick, comprehensive, and independent determination of which electronics are ideally suited to the customer’s intended application of AI. In this process, the researchers at ATKI consider the full range of electronics, from microcontrollers to powerful industrial computers, as well as all relevant performance parameters such as processing time, latencies, resource requirements, and power consumption.
Testing AI algorithms in a very small space
In addition, the scientists are investigating intelligent sensor systems in order to test their suitability and qualify them for industrial applications. Data reduction is a key aspect when it comes to the use of AI-based sensor hardware (edge AI). The researchers are analyzing various approaches to reducing data flow and evaluating their suitability for specific AI tasks, taking account of information loss and deriving recommendations for the handling of raw data and intermediate information.
Demonstrators enable practical tests of AI technologies and components
The range of services is completed by a demo production environment that the researchers plan to use for demonstrating the performance of AI technologies and components in a practical environment and for testing them in specific use cases. Using real machines and systems from the production environment, the researchers are therefore addressing subject areas such as condition and quality monitoring, data acquisition and analysis, AI hardware, and robotics in industrial transport and production processes.