Microsoft today announced that the next generation of HoloLens will have a Microsoft-designed intelligence chip, which will be used to analyze the visual data directly on the device, saving time by not uploading the data to the cloud service, giving the user a faster performance on the HoloLens 2 The device transfer feature as much as possible.
The announcement follows the trend of the largest technology companies now to meet the current artificial intelligence computational requirements. Current phones are not built to deal with these types of programs and on demand of current phones do so the result is a slow device or battery drain.
Artificial Intelligence works directly on handsets and enhanced reality goggles. It accelerates the performance of a device. There is no need to send data to external servers. It also makes the device more user-friendly because it does not require constant internet connection and is also safer because of data transmission. From the device to any other location.
There are two main ways to facilitate the existence of artificial intelligence on devices, first by building special lightweight networks that do not require much processing power, and second by manufacturing artificial intelligence processors, custom architectures and software, which is done by companies such as ARM and Qualcomm, Apple is building its own artificial intelligence processor for the iPhone called Apple Neural Engine, which Microsoft is doing now for the HoloLens.
This race to build artificial intelligence processors for phones works alongside the manufacture of specialized artificial intelligence chips for servers. Companies such as Intel, NVIDIA, Google and Microsoft are working on their own projects in this range.
Doug Borger, a research engineer at Microsoft, said the company faces the challenge of artificially building artificial intelligence processors for servers. “Their ambition is to have the first cloud service for artificial intelligence. Adding artificial intelligence capabilities to HoloLens can help achieve this goal, By focusing on the company’s expertise in chip architectures to deal with neural networks.
For the second generation of HoloLens, the artificial intelligence processor will be built into the HPU, which will process data from all sensors on the device, including the head tracking unit and infrared cameras; the artificial intelligence processor will be used to analyze this data using neural networks Deep, which is one of the main tools of artificial intelligence organization.
There is no official date yet for the release of HoloLens 2 but there are rumors of its release in 2019.