Fanless industrial AI system uses Nvidia’s Jetson

January 18, 2021 // By Nick Flaherty
Fanless industrial AI system uses Nvidia’s Jetson module
The Jetson-based fanless DLAP-301 nano, developed by Adlink, is now available through partner Acceed in Germany

Acceed in Germany has launched an industrial AI system based around Nvidia’s Jetson module.

The intelligent control and monitoring of processes, machines and systems is increasingly based on large data quantities and complex algorithms, whose calculation requires specialised hardware and a high bandwidth for data communication. In particular, the provision and processing of image data from connected cameras poses the controllers used in the industrial environment challenges. 

The fanless DLAP-301 nano, developed by partner Adlink, has a 64-bit quad core CPU ARM Cortex-A57 and an integrated 128-CUDA core graphics processor (GPU) with Maxwell architecture on the Jetson module. Combined with 4 GB 64-bit LPDDR4 memory, the module achieves 472 GFLOPS processor power. Nvidia specifies 25.6 GBit/s memory throughput. Here, depending on the mode activated, power consumption is only 5 to 10 W.

Eight RJ45 Ethernet interfaces with PoE serve for camera connection on the reverse side of the industrial casing with its width of a mere 21 cm. Moreover, two serial DB-9 sockets, three USB ports and a further GbE interface are available for data communication. Local graphics output is possible in high resolution via the HDMI-2.0 port on the front side.

The 12Vdc full-metal chassis measures 210 x 170 x 55 mm and is accessible from the front side, accommodates a 2.5” SATA SSD for local data storage. This means the DLAP-301 nano can simultaneously be used as an autonomous NVR (network video recorder), which can also stream image data to the Internet if required.

With the Jetson nano module, the DLAP-301 nano enables mature computer vision in real time and inferencing for several complex DNN (Deep Neural Network) models. For example, the control of multi-sensor autonomous robots could be realised or the integration of IoT devices with intelligent edge processing. Using ML frameworks, it should also be possible to train neuronal networks further directly.

The DLAP-301 nano is developed for industrial application under


Vous êtes certain ?

Si vous désactivez les cookies, vous ne pouvez plus naviguer sur le site.

Vous allez être rediriger vers Google.