Because one solution approach cannot cover all tasks, Adlink Technology has developed various AI products and a hardware optimization strategy in collaboration with chip manufacturers Nvidia and Intel. They support system architects in adapting edge computing systems to the respective requirements in terms of processing power, size, weight and power dissipation (SWaP).

Adlink's heterogeneous computing platforms consist of GPU- and VPU-accelerated products at the board, system and server levels, enabling system designers to construct and optimize architectures for both AI inference and training applications. By moving intelligence to the edge, heterogeneous platforms perform real-time streaming of data between edge devices and systems, ultimately leading to better decisions.

In addition to its computing products, Adlink offers consulting services using deep learning profiling to help users find the right platform to realize their applications cost-efficiently. The company also works with research and academic institutions to identify bottlenecks in AI platforms and analyze system issues that can be addressed to improve performance.

To illustrate the countless possible applications, Adlink will be showcasing a range of AI applications for the Edge on the stand:

• An automated optical inspection (AOI) application, based on an embedded computer and graphics module, finds solder defects on PCB assemblies under different lighting conditions.
• An automated counter reading system equipped with a deep learning accelerator recognizes and collects the readings of mechanical counters.
• A wood inspection application classifies samples by type with an AI-enabled smart camera and searches for scratches in the wood with an AI-enabled vision platform.
• An AI-accelerated frame grabber detects cracks in confectionery and reads deformed labels.

Adlink will also be showcasing its innovations for classically operated embedded applications.

Adlink Technology at embedded world 2019: Hall 1, Stand 540