WorldSense also provides an estimate of the last known position of the ground.
WorldSense uses advanced computer vision technology powered by dual fisheye cameras to precisely track the user's head position. It does this by combining feature points (visually distinct points in the environment) with inertial measurements from the headset's IMU to estimate the user's position.
WorldSense is designed to work in both dim and bright lighting, as well as both indoors and outdoors. As a rule of thumb, if the conditions are good enough for you to read a book, they're good enough for WorldSense to track accurately.
WorldSense can also work reliably even if one or both cameras are partially occluded.
WorldSense relies on feature points and textures in the environment to track. It can't track if you're looking at blank walls or other surfaces that lack sufficient texture detail.
When conditions are insufficient for WorldSense to track properly, it automatically falls back to 3DoF tracking. This might happen when there aren't enough feature points around, the lighting conditions are too dark, or the cameras are completely occluded.
If the conditions improve, WorldSense is able to completely recover and 'remember' where you are.
WorldSense uses an integrated neck model - developers do not need to implement their own.
Daydream Standalone automatically enables a safety fog to restrict tracking and keep the user in a smaller space. All end users will have the safety fog turned on. You can turn off the safety fog in developer mode for development purposes.
Worldsense uses a head-centric coordinate system. This makes it easy to port from 3.3 to 6.3 as it's the same coordinate system as in 3.3. When we recenter, we reset the origin. WorldSense also recenters each time the user puts the headset back on.