Focus 3 improving play area detection with specific wall images

[Frankfurt]

Hello, we are currently facing the issue that the tracking/play area gets lost too often. Since we are using lighthouse-stations and trackers in addition, it is a bit tedious to recalibrate the whole system over and over again. I'm assuming that the area will get detected via computer vision magic using the inbuild cams and I'm gessing that this is not a trivial task (for example setting up the play area in evening in my office (lit by lamps) and finding that exact same positions / tracking features under daylight conditions) So to make this more stable I'm thinking of placing black and white marker images on the walls (something equal to AruCo markers) that will be more or less similar under all lighting conditions. Can you confirm can this would increase the stability? And if so would my approach using such kind of aruco images help the best or would you recommend other types of images? Thank you very much!     

Edited April 14, 2022 by Frankfurt

[HackPerception]

@Frankfurt

• For lighting conditions - I would recommend just using the pass through cameras and figuring out if there's any obvious places with over exposures or blowout.
  • Use the passtrhough to ensure there's a decent array of high-contrast tracking points spread out throughout the environment.
• Yes, auCuo markers will help. You probably want them to be at least 12cm across.
• You can also use tape and other pretty much anything that's high contrast to create additional "textures" for the headset to track.

Focus 3 actually has some more advance arUco support offered as part of a location customization service. It allows you to virtually align a virtual scene to the real world via arUCo.

 

[Ixixly]

With the arUCo symbols, the information I've seen has me under the impression that they are more Application specific rather than System specific, as in developers need to include support for them and for specific ones rather than being able to put them around an area for the Headset itself to recognise? Could you elaborate a bit more for us on this?

[Frankfurt]

@HackPerception ... We are going to do some first "stresstests" quite soon so these are exellent suggestions and will (hopefully) help a lot! Thank you very much! 

[o.pak]

Hey guys🖐️
- could you please tell me, how the process of integrating #arUCo into your products is going?
- to what extent do you feel arUCo stabilize the tracking of one(or more HMDs), and in what area?
- also, maybe you check how much more the helmet's GPU loads when arUCo recognition is enabled? Thx!