Skip to main content
April 4, 2014
Question

Positional Tracking for vast open spaces?

Hi People

I'm currently doing some research on positional tracking. I didn't know if there were any people on the forum who really have extensive knowledge on the matter? What it is I'm actually looking for is a positional tracking package that can work in vast open spaces on a mobile devise to allow accuracy of a VR/AR image it's one area of my project I'm working on that isn't essentially needed but would be a great added bonus to be able to include, any info would be greatly appreciated

Thank you

    23 replies

    Up and Coming
    April 4, 2014
    GPS?
    Currently working on Dolphin VR Follow along: https://github.com/CarlKenner/dolphin/commits/VR-Hydra Latest Version: viewtopic.php?f=42&t=11241&start=1020#p249426
    April 4, 2014
    Needs to be a bit more precise! (Maybe it's called local)
    Also does not have to be as distant as GPS, however GPS would work if the tracking was precise enough to gain a 1mm tolerance!
    It needs to track a users position at certain co-ordinate and then display the image as near to precisely from a point of reference

    Also I'll just add I don't think the tracking needs to be picked up from a satellite!

    As there are predetermined engineering marks already in play in construction I believe the the point of reference needs to be picked up/read from a couple/few of these engineering marks as all the the dimensions and measurements are already accounted for from the plans, then from that point the users location needs to be read from say lasers positioned around the user?

    I will add I know nothing about positional tracking other than what I have read over the last day!

    So if none of this makes sense please forgive me however maybe somebody could elaborate and point me in the right direction
    Up and Coming
    April 4, 2014
    Currently the only method that can get 1mm accuracy is optical mocap. So for large area you want the kind of Mocap Setups they use to make films like Avatar, Lord of the Rings etc. With high def CCDs getting cheaper and using IR LEDs instead of those expensive reflectors it might be possible to set up a Mocap studio more cheaply these days.
    April 4, 2014
    Thanks does my theory make sense if you're up on the positional tracking?

    This is how I see it in construction there is already a virtual mapped out design made of equations and numbers that are determined from pre-determined marker points/engineering marks, from what I've said previously do you think it would be possible for the the software to gain point of reference from the pre-determined engineering marks already present in construction? Then the user could be determined by a series of lasers?

    My point I'm trying to get to is that if the pre-determined engineering marks are sufficient to map the outline of the image accurately then surely if that info is passed on from the device itself surely the lasers that track the user could determine the correct point of reference from the layout and engineering points of the plans themselves ?

    Just theory but I'm pretty sure it could work and there would be no point to using GPS etc?
    Up and Coming
    April 5, 2014
    I dont know exactly what you are doing but t sounds like what you probably want is QR codes (and ignore my post about infrared mocap). Valve has successfully used QR codes in VR demonstrations for motion tracking. QR codes carry data like a bar code so you have the option of writing the co ordinates into the QR code or just a serial number for use with a database. If a camera can see the corners of the QR code then as the QR code is of known size and dimensions a cellphone can tell its 3d position and orientation relative to the QR code. Its a simple triangulation calculation from the four corners. So if a cell phone can see at least one QR code with a known fixed position it knows exactly where it is. With some clever programming it will be able to tell which square is which/where before it is close enough to read it by its position relative to the other known qr codes.
    April 5, 2014
    Ok I've designed a new system for the construction industry using AR I will also be using VR with the system but the VR side of things will only be used from indoors in the office space, So far the current device will use a Visor attached to a hard hat allowing ease between both scenarios (AR and Reality).

    Basically the system will offer a complete guide to steel fixing and then later on the software package will run throughout the board of construction.

    From the visor the user will see (if positional tracking is used) a complete guide of how to carry out the works they will also see a complete 3D Augmented image of the plans, as if its already there, the positional tracking isn't 100% needed but would be good to add especially for future proofing.

    Do you think while outside instead of say using the QR codes, lasers on tripods would be able to take their place to give the position of the user?
    Up and Coming
    April 7, 2014
    When you say lasers on tripods do you mean surveyor type lasers pointing a the hard hat to measure distance? Surveyor lasers typically measure to within a few millimeters, but in order to keep pointed at the hard hat it would need to have some motorized automatic tracking, if you knowing the angles from the tripods you can triangulate from the angles anyway making the lasers superfluous. Which brings us back to optical tracking. And array of infrared LEDs like on the DK2/Crystal cove tracked by near infrared camera on tripods, to work accurately from long distance they would have to have motorized tracking from a zoom lens. A construction site is a tricky environment with new walls and stuff going up blocking line of site. But the like the QR codes the LED dots on the hard hat are of known dimensions so as long as one camera keep track of you the angle, orientation and distance to the camera can be calculated. But this would not scale well to multiple users as using wide angle lenses to track multiple users would reduce accuracy, Oculus recommends for using the DK2 within 2.5 meters of the the camera for instance.

    This could also be done by putting QR codes on the helmet, the reasons Oculus went for an LED pattern DK2 instead is that they can put a narrow frequency infrared filter on the cameras CCD. So that the LEDs are pretty much all the camera can see making recognizing them easier, it can also then use 1bit color allowing it to to run faster at a higher resolution. Apart from the infrared filter its just a standard camera so pretty cheap. LEDs can also be turned off and on in sequence helping the software sort out which is which. LEDs are also cheaper than the infrared reflectors used by the film industry.

    They parts can be had off the shelf but it would probably require allot of custom programming even if you licensed existing MoCap software.

    The reason I suggested QR codes before is that I though it sounded like a VR or AR experience in an exhibition type space that might have lots of people walking around with mobile part driven VR helmets, which would suit itself to the Valve solution of covering the walls and floor with QR codes.

    There are lots of industries that would like millimeter accurate position/motion capture without wires or line of site but unfortunately there is no magic bullet solution.
    April 7, 2014
    Yes that's right about the surveyors lasers typically most of them are motorised anyway. They continually spin as they run off a chargeable battery they calculate distance from the time break in the laser from the time it hits the designated devise e.g a staff. In this case would be the hard hat or the devise attached to it the user is wearing.

    They also need to be able to be moved around freely which is why I still believe the point of reference will need to be taken from the dimensions provided by the plans.(still thinking about this one)

    Again the 1mm tolerance would be very handy but not essentially needed I'm thinking more along the lines of, if weather or not to include positional tracking in the prototype model, in an ideal world I would prefer to add it.

    Especially as my vision for the future product will definitely have it included

    I'm pretty certain it's possible may not be as accurate as 1mm though bit would be great to add the feature
    April 8, 2014
    Something's shouting out at me that drones could well be used to gain the positional tracking needed?
    Up and Coming
    April 9, 2014
    But the like the QR codes the LED dots on the hard hat are of known dimensions so as long as one camera keep track of you the angle, orientation and distance to the camera can be calculated.


    I should correct myself here, even with a zoom lens the depth data from a single camera as triangulating from width of the head at distance small angles lead to larger movements. So two cameras would be needed for that method at range.

    A laser would of course give very good depth data from allowing full position tracking from a single tripod. But I would expect difficulties when adapting surveyor instruments as they where not designed for tracking moving targets.

    Just saw this video on another thread. Radio based positional tracking I am very impressed.

    https://www.youtube.com/watch?v=mYyFUQbWC1E