Intel aims to improve IoT products with AI

Intel has teased the fruits of its work into making smart home products and other network edge devices more ‘intelligent’ with the announcement of its Neural Compute Stick 2, ahead of its inaugural AI developers conference in China.

A ‘neural network on a stick’ may not be catchy, but the potential for the development of current and future products for the smart home could be significant.

To use its full name, the ‘Neural Compute Stick 2’ could be a great help for developers of future connected products for the home thanks to supporting the development of AI algorithms and computer vision systems locally without needing the Cloud (or the internet at all).

A mere Linux-based OS with a USB 3.0 port is all that’s needed to work with the NCS2, which essentially can take on all the processing work from other products (thanks to its Movidius Myriad X vision processing unit) and speed up and simplify their development.

Neural Compute Stick 2 front view

The big difference between NCS2 and its 2017-released predecessor? Performing as much as 8 times faster when it comes to processing, says Intel.

“We can do some incredible things with artificial intelligence, and one of the fastest growing data types, as we all know, is video data,” commented Steen Graham, general manager of Channels and Ecosystem in Intel's Internet of Things group, on its launch. He added: “The camera is the ultimate sensor.”

A small number of developers have already been working with it on a number projects including Clean Water AI & BlueScan AI, but the NSC2 could potentially be used across a broad range of products (where data being transferred between two networks) – such as routers, switches, gateways and connected doorbells. Of course, it could also lead the way for a number of new products and features coming into the home too via IoT and edge devices.

The Neural Compute Stick is now available to everyone for $100 from Intel.