DIY Head Tracker For A Tenner

reaper7

reaper7
n0ahg said:
The MPU-6050 already is 6DoF. You can then fuse the magnetometer in software as above but you'd still only get orientation!
Click to expand...

Should have been clearer with my question, indeed hardware is 6DOF does the software EDTracker support 6dof or is it only 2dof.
While elite dangerous only gives us up/down left/right, can I use this software with other sims to have full 6dof of axis?
 

brumster

brumster
EDTracker only does 3 "true" DOF (forget all the hardware marketing hyperbole of 6/9/10DOF etc).

Yaw
Pitch
Roll

As you say, elite doesn't make use of the Roll for headlook, but the device does report it on a 3rd joystick axis, should you want to use it. Try it for sideslipping in docking mode ;)
 

n0ahg

N
brumster said:
EDTracker only does 3 "true" DOF (forget all the hardware marketing hyperbole of 6/9/10DOF etc).

Yaw
Pitch
Roll

As you say, elite doesn't make use of the Roll for headlook, but the device does report it on a 3rd joystick axis, should you want to use it. Try it for sideslipping in docking mode ;)
Click to expand...

Spot on. This needs to be in a FAQ somewhere :->

In mechanics 6DoF is orientation and position. In sensor terms 6DoF simply means you have 6 data inputs to work with (this is why you have 9DoF when you add a magnetometer!)

As position is the second integral of acceleration you'd get a lot of errors which is why Oculus added optical tracking for position and STEM uses magnetic fields for accurate position and orientation. I have a STEM system on the way :)
 
Last edited:

reaper7

reaper7
What about forward/backwards movements? Does software allow for this.
Looking to replace a TrackIR device that has roll, yaw, pitch and forward axis.
Would de great if this did the full range if motion for sims that support it like dcs and falcon.
Find teackIR has very erratic response at the best if times.
 

brumster

brumster
As n0ahg says above, I would give up hope of getting any lateral/forwards/backwards movement represented out of an MPU-only device (ie. one with only accelerometer/gyro/magnetometer)... at least accurately, anyway.

You could probably detect G-forces forwards/backwards and have the device "step" into two or three positional zones; so monitor for a sudden "nod" or "jerk" of the head forward or backwards, and report this as a change in another axis that could be used to represent zooming of some sort. But it would be a bit clunky and horrible, and prone to erroneous actions should you accidentally jerk your head around too much!

The whole 6DOF thing isn't really something we've been pitching EDTracker as. The intention was cheap, effective head look (3DOF). I believe this gives you 90% of the benefit in first-person games like Elite, Arma, DCS, Truck Simulator, etc. If you want that extra 10%, I would look towards TrackIR or one of the point-model based systems like FreeTrack/Opentrack. Or one of these STEM systems.... :)
 

reaper7

reaper7
Ok thanks guys, that clears it up for me.
Will keep the TrackIR for the flight sims and build this for now for ED and star citizen.
At least untill CastAR is available, that will be the ultimate solution for cockpit builds.
 

Antmax

Antmax
CaptainSodom said:
I don't see much point in roll in headlook mode, although I imagine it would make for a bit more realistic-feeling experience.
Click to expand...

3DOF does make some difference. Having come from Opentrack to EDTracker. 3dof is more natural when tracking. It doesn't sound it but when you can roll your head to match your quarry it makes banking the ship and following slightly easier. Especially when it is crossing your up into diagonal corner.

Also you do not need to move your head as much with 3dof. You can move your head about 30% less and then nod sideways down (roll head) to open up one of the computer displays while still being able to see much of the scanner. Same with tracking as I touched on in the first paragraph of my post.

So definitely worth trying to attain.

I managed to get 3DOF with EDTracker by using opentrack and a utility that uses Vjoy to translate the EDTracker joystick data to something opentrack can understand. Unfortunately the long convoluted path or EDtracker, Joystick curves, vjoy, opentrack does have an impact on precision. So I'm using standard 2DOF for now.

Pocket moon mentioned elsewhere that he thinks he knows why EDTracker joystick axis isn't recognized by Opentrack joystick input plugin. Something to do with the HID descriptor or something. He aims to fix that at some point so we have more options (like almost any TrackIR compatible game working in 3DOF with EDTracker)
 

mojonaut

mojonaut
Drift to the right

Having an odd issue with drift, Tracker is virtually drift free, if clamped to bench for an hour it hardly moves off center.

If I play the game as long as I am in Headlook mode it again remains centered. If I switch out of Headlook for any length of time when I go back to Headlock the Tracker center viewpoint has moved fully to the right and only the reset button will restore it.

Is it a known issue?

thanks
 

NeilF

NeilF
mojonaut said:
Having an odd issue with drift, Tracker is virtually drift free, if clamped to bench for an hour it hardly moves off center.

If I play the game as long as I am in Headlook mode it again remains centered. If I switch out of Headlook for any length of time when I go back to Headlock the Tracker center viewpoint has moved fully to the right and only the reset button will restore it.

Is it a known issue?

thanks
Click to expand...
Recently (the last couple of times I've played for 1-3hrs) it has been almost rock solid for me...

Before this, I was having to reset every 30mins or so, and quite frequently when coming out of Starport services, when, like you, the view would suddenly be skewed a long way right/left.

But as I said, the last couple of times I've played ED, the tracker has been practically perfect!
 

Antmax

Antmax
I get this drift problem too. I was wondering if it was the auto centering. I was reading on how if you keep your head off center for too long the software resets the center. Over a long time it's possible with a lot of recurrences it might be drifting off center.

So when you put it aside for a long period the center might have drifted trying to auto compensate. Pressing the reset button then puts it back to true 0,0,0.

Just speculation, but might explain the behaviour you are experiencing.
 

NeilF

NeilF
Antmax said:
I get this drift problem too. I was wondering if it was the auto centering. I was reading on how if you keep your head off center for too long the software resets the center. Over a long time it's possible with a lot of recurrences it might be drifting off center.

So when you put it aside for a long period the center might have drifted trying to auto compensate. Pressing the reset button then puts it back to true 0,0,0.

Just speculation, but might explain the behaviour you are experiencing.
Click to expand...

I believe this "auto centering" is only done when you are looking for a period of time near the existing center. It assumes you are looking forwards, but it has drifted so resets itself. But again, I believe this is only when looking slightly off center, not when a long way off.

That's my understanding. But I gave no idea of what the margins are, or the time frame.
 

brumster

brumster
Yes, exactly. The auto-recentering just works by monitoring where you are looking, and if you are *just off* dead centre by a small amount, it assumes the reason you're slightly off is because of drift, and tugs the view back to centre. This works because, basically, no-one ever really looks left or right by a fraction of a degree - in game, you're looking around you significantly (ie. over to one side to track a target, or up).

So the EDTracker guesses that any time it's pointing slightly off-centre, your head is actually dead centre and it's drifted - so it auto-corrects.

You can fool it yourself by looking ever so slightly to the left/right and holding your head there for a fraction of a second, and it will auto-centre to the new position.
 

mojonaut

mojonaut
Antmax said:
I get this drift problem too. I was wondering if it was the auto centering. I was reading on how if you keep your head off center for too long the software resets the center. Over a long time it's possible with a lot of recurrences it might be drifting off center.
Click to expand...


Yes, I remember mention about the small resets while looking forward.

But I'm finding hardly any drift if the unit is left clamped to the desktop, only when in use and not in headlock mode in ED: does it seem to drift, which makes no sense.

The only other thing I can come up with is it is warming up when sat on the top of your head?


I'll try some more tests with the monitoring software.
 

NeilF

NeilF
brumster said:
Yes, exactly. The auto-recentering just works by monitoring where you are looking, and if you are *just off* dead centre by a small amount, it assumes the reason you're slightly off is because of drift, and tugs the view back to centre. This works because, basically, no-one ever really looks left or right by a fraction of a degree - in game, you're looking around you significantly (ie. over to one side to track a target, or up).

So the EDTracker guesses that any time it's pointing slightly off-centre, your head is actually dead centre and it's drifted - so it auto-corrects.

You can fool it yourself by looking ever so slightly to the left/right and holding your head there for a fraction of a second, and it will auto-centre to the new position.
Click to expand...

Is it really just "a fraction of a second" the test/adjustment is made over?
 

NeilF

NeilF
brumster said:
It monitors it over 10 loops through the main routines - each loop taking just milliseconds - and then applies the correction every second.
Click to expand...

Seems a very short time to build up an impression? But I'm sure it's been checked/tested countless times :)
 

Shorn

Shorn
Hi folks,

I _think_ the Starport services issue is linked to the auto-correct. A week or so ago I did some tests as I was getting a bit peeved with the issue of resetting every time I came out of starport services to find the view massively skewed left or right (yes, yes, I know - small problems take on a disproportionately large significance when they are pretty much the only problem!!!)

I believe when in starport services my head turned slightly to the left or right (e.g. when in commodity market I was looking a bit more to the left, reading through the items while making a choice, or in outfitting looking slightly to the right while viewing the available upgrades) just by the position of the items onscreen, and there is no obvious central point of reference.

When I made a conscious effort to keep my head position "locked" to the centre of the screen the skew on exit was reduced significantly, sometimes gone completely.

When out of the the starport screen I could "force" it to autoadjust sometimes, looking slightly left or right and could see the screen move the opposite way. But sometimes I couldn't get it to shift at all (that may very well have been my inability to find the central point to look slightly away from, as once the view has moved there is again no central reference point!)

Just a bit of information really, if anyone wants to try and replicate or if it helps alleviate someone else's peevishness :)

I have to say my "bit peeved" is about the only issue I've had with this fantastic tool, 1st world problems 'n all that.....

Shorn
 

Lasse B.

Lasse B.
I'm getting a couple of errors again when using the newest sketches with IDE 1.5.7:

Code: 
EDTrackerII.ino:10:21: error: variable 'infoString' must be const in order to be put into read-only section by means of '__attribute__((progmem))'
EDTrackerII.ino:184:1: error: 'TrackState_t' does not name a type
EDTrackerII.ino: In function 'void loop()':
EDTrackerII.ino:492:9: error: 'joySt' was not declared in this scope
EDTrackerII.ino:498:9: error: 'joySt' was not declared in this scope
EDTrackerII.ino:516:7: error: 'Tracker' was not declared in this scope
EDTrackerII.ino:516:25: error: 'joySt' was not declared in this scope

and

Code: 
EDTrackerIICalibrate.ino:5:21: error: variable 'infoString' must be const in order to be put into read-only section by means of '__attribute__((progmem))'
 
You must log in or register to reply here.
Back
Top Bottom