Skip to main content

aryve Vision

Foo diagram Foo diagram

What is it​

The aryve Vision component is a camera-based positioning module that enables high-precision indoor localization using visual input from a device.

The component processes live camera data, matches it against a pre-mapped environment, and returns the device’s precise position and orientation within the aryve coordinate system.

What is it used for​

The Vision component is designed for:

  • High-accuracy indoor positioning in pre-mapped spaces
  • Visual localization without hardware installations
  • Navigation in any indoor multi floor space
  • Augmented Reality (AR) navigation experiences
  • Robotics or device guidance indoors

How it integrates into the aryve ecosystem​

The aryve Vision component integrates as a positioning service within the aryve ecosystem, providing real-time, camera-based localization to client applications and other aryve services.

The Vision component relies on precomputed visual maps generated within the aryve ecosystem. At runtime, the Vision component continuously matches incoming camera frames against this map data to determine the device pose and centimeter precise position. The system also keeps the information of the device's session to increase the precision based on the navigation context of the device.

aryve Vision communicates with the aryve Studio and aryve Guide components via REST API.

Input requirements​

What is needed to use aryve Vision is a device with a camera and access to internet, that can send the image to the server to be localized and receive back the position of the device.

The camera requirements are:

  • RGB camera
  • Minimum resolution: 1024x768 pixels
  • Camera intrinsics

Device requirements:

  • Image processing (jpeg encoding)
  • Access to camera stream
  • Network connection

Physical requirements recommendations​

  • At a height of 1,5 meters the camera should be pointing horizontally
  • If the camera is lower/higher then it has to be adjusted to not point at the floor or ceiling but at the surrounding space
  • While we can work with some motion blur, fast rotation and shakiness should be avoided

Output data​

The Vision component returns real-time localization data to the client device.

  1. Position
  • x, y, z coordinates
  • Defined in the map global system (every map has an origin point and every position in the map is referencing to it)
  • Unit: meters
  • With orientation
  1. Accuracy
  • Position accuracy estimate
  • Localization status (failed, not failed)

Important to note​

  • Performance depends on lighting conditions, clarity of the image, visual features richness (Note:standard smartphones are better at handling light conditions)
  • Map quality might degrate over time in a changing environment
  • On devices not using the aryve Guide app, people have to be blurred on the device to ensure you are GDPR compliant