track user translation movement on ios using sensors for vr game?

Question

I'm starting to experiment VR game development on ios. I learned a lot from google cardboard sdk. It can track user's head orientation, but it can not track user's translation. This shortage cause the use can only look at the virtual environment from a fix location (I known I can add auto walk to the game, but it's just not the same).

I'm searching around the internet, some says translation tracking just can't be done by using sensors, but it seems combining magnetometer, you can track user's movement path, like this example .

I also found a different method called SLAM, which use camera and opencv to do some feature tracking, then use feature point informations to calculate translation. Here's some example from 13th Lab . And google has a Tango Project which is more advanced, but it require hardware support.

I'm quite new to this kind of topic, so I 'm wondering, if I want to track not only the head orientation but also the head(or body) translation movement in my game, which method should I choose. SLAM seems pretty good, but it's also pretty difficult, and I think it will has a big impact on the cpu.

If you are familiar with this topic, please give some advice, thanks in advance!

Answer 1

If high accuracy is not important, you can try using the accelerometer to detect walking movement (basically a pedometer) and multiply it with an average human step width. Direction can be determined by the compass / magnetometer.

High accuracy tracking would likely require complex algorithms such as SLAM, though many such algorithms have already been implemented in VR libraries such as Vuforia or Kudan

Answer 2

Hi I disagree with you Zhiquiang Li Look at this video made with kudan, the video is quite stable and moreover my smartphone is a quite old phone. https://youtu.be/_7zctFw-O0Y