Rokid Vision

On May 21, Shanghai HiSilicon officially released the XR chip platform. At the same time, it launched the first XR chip that can support 8K decoding capability. This chip comes with an integrated GPU and NPU. The first AR glasses based on this platform are the Rokid Vision.

hiSilicon XR

Huawei HiSilicon XR

Hisilicon has also made more interpretations on this platform:

  • The XR chip platform supports 8K hard decoding capability and can support up to 42.7PDD per eye.
  • Taking VR as an example, the ideal resolution that the human eye can achieve is approximately 60PPD (pixels per degree). Generally, a chip that supports 4KP60 decoding capability can support a single-eye resolution of 21.3PPD, which corresponds to 480P. The TV display effect is far from being comfortable and immersive.
  • The XR chip released by HiSilicon also improved the decoding ability to provide a clearer content rendering effect.
  • The platform joins HiSilicon’s proprietary architecture NPU, which can provide up to 9TOPS of NPU computing power.
  • An important application scenario of AR products is face recognition and license plate recognition. This requires real-time analysis of the video images collected in the AR glasses camera, correctly identifying the face and license plate in each frame of the image, and uploading the recognition results to the server. There, it performs the comparison and displays the comparison result on the screen in the glasses in real time.
  • From image acquisition, analysis, upload to display, the shorter the time. In this sense, this product can achieve hundreds of people faces and license plates complete analysis per minute and recognition in milliseconds. This requires the chip to have a strong NPU computing power support.

Based on such considerations, HiSilicon has optimized the platform architecture.

Editor’s Pick: Apple took 66% of smartphone industry profit: how the Chinese smartphone brands hit the high-end market?

Rokid Vision Features

As the first product of HiSilicon’s XR chip platform, the Rokid Vision is a core product of Rokid, a domestic AR startup.

The front appearance of this product is similar to ordinary sunglasses. There are two independent displays on the inside of the lens, with 40°+ field of view (FOV), 1080P 3D display effect, optical waveguide display option, and stereo sound on both sides of the temple. So it can bring users an immersive experience equivalent to a 110-inch 3D HD TV 4 meters away.

It is reported that the Rokid Vision will be used in security, industry, education, retail, exhibitions and other fields to provide a more powerful, portable, and immersive experience.

Rokid also officially announced its joining of HiSilicon’s XR chip platform. Prior to this, Rokid has optimized this platform-specific algorithm and program together with HiSilicon.

The two parties have done a lot of optimization work on AI computing capabilities such as voice, vision, and environmental perception. They have completed the high-resolution and low-latency design of multi-channel video encoding and decoding.

Argam Artashyan

By Argam Artashyan

Back in 2010, he was dismissed from his position as a lecturer at the university. This made him get another job at his friend’s digital marketing company as a blog writer. After a few years, when he was thinking the article writing is his mission, Google pushed the Panda update and affected the company and websites he was working at. (Un)fortunately and surprisingly, he got an offer to head a large knitting factory. In 2016, he got his Ph.D. and resumed teaching at the University … and writing tech-related articles following his passion.