In this video, we will continue talking about VR hardware.
In particular, we will focus on head tracking.
First of all, as mentioned in the previous videos,
we tracked the user's head in order to
update the displays according to the user's viewpoint.
It's important we can do it very quickly to maintain emotion.
Most VR devices on the market have
this tracking latency controlled at level of single digit milliseconds.
So it's not noticeable in most cases.
There are other processes in this procedure that could
cause further delays in updating the graphics,
but they are independent of the tracking technology we're talking about here.
And we will discuss those other factors later in the course.
Secondly, we need to track the user's viewpoint in the 3D space.
This implies we have to track their head rotation and position.
In the CAVE, there are normally built-in systems in
the small tracker which the user wears on top of their shutter glasses.
The built-in tracking system tracks
the user's head rotation using accelerometer or gyrometer or both.
They can be very tiny and the same methods are actually used in
most smartphones in order to support
features such as autorotation when you flip your phone like this.
For position tracking, external optical tracking devices are normally used.
In the CAVE, it could be several ceiling-mounted infrared cameras pointing at
a user so they can calculate the position of the user in
the 3D space and feed the data to the machine,
which then takes care of updating the display accordingly.
Here, how you mount the cameras is important as this defines the range of capturing.
So when you come outside of the CAVE or come very close to one of the walls,
you might lose position tracking as not all cameras can see the head tracker.
The same applies to HMD head-tracking.
We would need to use built-in accelerometers and gyrometers to track
the user's head rotation and external optical devices to track their head position.
In this case, the external infrared sensor
is mounted in front of the user as you can see here.
And again, the range of capturing depends on how the sensor is set up.
If the sensor is set up in front of you and you move
away to a place where the sensor can no longer see you,
then we lose position tracking,
which means the graphics can no longer update according to your viewing position.
It is important to mention that not all devices come with position tracking.
Some have rotation tracking only.
This means that you can still look around but you are stuck where you are.
You won't be able to move around in VR with your body to get closer to
certain objects or observe them from a different angle like you can in real life.
You can still navigate the 3D environment with a controller,
a touchpad or joystick,
but it won't be as natural,
and you won't have a good sense of scale in VR without this position tracking.
In fact, none of the mobile devices currently on the market come with position tracking,
be it Google Daydream or Samsung Gear VR.
It's normally a HMD box for you to slot in your smartphone,
and they don't come with
external optical tracking sensors like the high-end HMD devices do.
Part of the reason for this is that mobile VR
is very much about their unsophisticated setup,
and that you can move freely in a room without out of range problems.
So it's against the design principles to add
a sensor which will then bring new constraints.
Also, integrating position tracking and updating the graphics in
real time potentially could consume a lot of computational power,
which is more difficult to support with just a mobile phone.
There are possible technical solutions to track the position of the HMDs
without relying on external position tracking devices.
For instance, a technology called
self-tracking or inside-out tracking has had some success.
It is beyond the scope of this course but you're free to read about it.
It's basically still a challenging problem in machine vision.
You might get it to work in certain conditions,
but it might not be robust enough to work in
everyone's living room with their own personal lighting conditions.
At the time of recording,
I have not seen a consumer-oriented VR headset that comes with robust position tracking,
but this might change quite quickly over time.
It is something to watch out for before buying your VR device.
It is possible in the near future we will have wireless HMDs
powered by desktop machines, for instance, Bluetooth.
So users are free from the constraints of
cables yet benefiting from a very powerful machine,
which enables sophisticated graphics and
interactions beyond the computational power of a mobile phone.
Dr Sylvia Xueni PanLecturer, Department of Computing
Dr Marco GilliesSenior Lecturer
