Kontron, a leading global provider of IoT/Embedded Computing Technology (ECT), is now supporting companies with a compact platform for Artificial Intelligence, Machine Learning and Deep Learning. It integrates all Google TensorFlow Lite applications and enables rapid development of custom AI-applications. The Google Coral Edge TPU (Tensor Processing Acceleration Unit) with up to 4TOPS (trillion operations per second) ensures higher speed in image and video data processing. Compared to an application with simple USB-cameras without TPU at approx. 6 frames/s, the TPU accelerates to a speed of 30 frames/s, so five times faster.
Artificial Intelligence in the Intelligent Edge is gaining more and more importance in industrial automation. The new compact Kontron AI platform consists of an M.2 module with the Google Coral accelerator chip for the software ecosystem TensorFlow Lite on a 2.5″ pITX SBC from Kontron with an NXP i.MX8M processor. The TPU supports small and low power applications and requires only 1 Watt for 2 TOPS, so 2 Watt for 4 TOPS.
All current applications of TensorFlow Lite as well as pre-trained models can be downloaded free of charge. This allows a fast and uncomplicated development of own neural networks with AI/ML/DL applications, and thus a significantly reduced time-to-market.
In the context of Industry 4.0 applications, topics such as object recognition and classification as well as quality inspection of objects and predictive maintenance, which are based on AI-algorithms, are increasingly being addressed. Artificial Intelligence is also playing an increasingly important role in point-of-sales applications in order to place advertising and relevant information in a more targeted manner. Kontron’s solution can also be flexibly customized to meet specific application needs. The platform is suitable for use in harsh environments with extended temperatures ranging from -40° to + 85° Celsius and supports Yocto Linux with the latest kernel. The platform will also be available on other Kontron systems within the first quarter of 2021.