Skip to main content
rldai
Up and Coming
May 27, 2013

Oculus VR with motion capture

{update20140809}
hi guys, we put it on Kickstarter.We worked really hard to make it affordable and easy to use. Now you guys are able to get this kind of stuff home and play with it. We are exhibiting a few days later at Siggragh 2014, booth 533.
Back us :)

https://www.kickstarter.com/projects/1663270989/project-perception-neuron?ref=discovery




{update20131025}


1st Rift/trueMotion MoCap/FPS demo


[updata20130829]
Successfully combined our motion capture system with Oculus Rift.
Now you can walk and run in the virtual world, and you can touch things by using your own hands like this!
Oculus with trueMotion2.jpg
Oculus with trueMotion.jpg



some recent update on VR with motion capture to share here (20130714)
It's in Chinese but I believe you'll understand what I mean
3 screens from left to right
1, from your goggle, 2. 3rd party view, 3. real-time captured video
http://youtu.be/EGZ5Y5x95P4

and... we are flying to Anaheim, California for Siggragh 2013 exhibition next weekend and the week following. Come and check us out!
http://youtu.be/R6pnPggN0Tk



Hi guys,

Here is a motion capture system developed by my team using IMUs (demo videos list below).

0. Peking opera realtime captured
http://www.youtube.com/watch?v=Gul7LWna3KI
1. Fullbody motion capture
http://youtu.be/DPGEG8U7Be8
2. Lowerbody motion capture
http://youtu.be/QgyY3aRtiwk

Those are real-time captured by using match-box-sized wireless IMU sensors.

I'm considering whether it is possible to combine our system with Oculus Rift which will make an idea virtual reality environment.

I’ll start to make Oculus Rift work with my motion capture system as soon as I get the dev kit. For a full body motion capture we need 17 sensors but my thought is that lower body tracking (7 sensors) is good enough for this. Maybe one more for chest?

Although my motion capture system is fully wireless, I think I still need something mobile to feed the video to Oculus Rift. Does anyone know how to transfer DVI video wirelessly? I really don’t wanna carry a 5 pound IBM laptop while running around in my office.

    28 replies

    Show previous replies
    geekmaster
    Up and Coming
    June 17, 2013
    "brantlew" wrote:
    ... Kinematic constraints offers a neat little "cheat" to this limitation but still requires a number of sensors to work well. ... bucking the conventional wisdom ...

    Which is why I am using kinematic (and skeletal) constraints, including the user's buttox [NSFW] constrained to his (or her) chair. You can determine in which direction and how hard a nudge (jerk) was applied, and use that to estimate a speed and resting position, providing that the user cooperates with the algorithm and uses it to best effect. Much better than just IGNORING head position changes, IMHO. And my algorithm also detects tap and double-tap of my fingers to the left and right sides of the Rift, to simulate button presses (for menu controller-free menu navigation). In fact, it is sensitive enough to detect footsteps on the floor in the next room, when my Rift rests on my desk. I need to divert more "forum time" back into my code, but especially this particular code.
    :D

    And of course, there is little left to be discovered within "conventional" wisdom. TRUE wisdom seeks novel ideas and new solutions outside the confines of the convention box. You will not discover new tools if you limit your search to the same old toolbox. We all need to start thinking like pioneers here, which requires a lot of "seat of the pants" tinkering, and a dash of "yankee ingenuity". IMHO.
    ;)
    This work is licensed under a Creative Commons Attribution-ShareAlike 3.0 Unported License.
    rldaiAuthor
    Up and Coming
    June 18, 2013
    "brantlew" wrote:
    The videos are really impressive - especially the second video. Your physics model for handling flight seems very good and the motion capture seems to approach the quality of optical mo-cap. Thanks for sharing.


    Thanks! Yes I developed even more equations than my PhD study in the past 2 years ... :geek:
    rldaiAuthor
    Up and Coming
    June 18, 2013
    "geekmaster" wrote:
    The developers must have ignored the naysayers who claim that positional motion capture with IMUs is impossible (or at least not good enough for head tracking).


    They are right about IMU only gives posture readings but you can go further to build up a physical engine and human body kinematics model to get the position.
    rldaiAuthor
    Up and Coming
    June 18, 2013
    "brantlew" wrote:
    "geekmaster" wrote:
    The developers must have ignored the naysayers who claim that positional motion capture with IMUs is impossible (or at least not good enough for head tracking).


    Well generally I think the naysayers mean that position for a single sensor cannot be determined. Kinematic constraints offers a neat little "cheat" to this limitation but still requires a number of sensors to work well.

    I am happy to see people working on this though - bucking the conventional wisdom about optical mo-cap. IMU mo-cap could have major benefits in terms of the variety and scale of environments.



    And you can squeeze a full set moCap system into a lock-lock box.
    More important... it's seriously much cheaper...
    geekmaster
    Up and Coming
    June 18, 2013
    "rldai" wrote:
    ... And you can squeeze a full set moCap system into a lock-lock box. ...

    What's a "lock-lock box"? Google isn't helping me on that one... :shock:
    This work is licensed under a Creative Commons Attribution-ShareAlike 3.0 Unported License.
    blazespinnaker
    Up and Coming
    June 19, 2013
    Very cool. Do you have any examples with two or more people?

    Also, do you use it in conjunction with a camera as well?
    rldaiAuthor
    Up and Coming
    June 24, 2013
    "rldai" wrote:
    "brantlew" wrote:
    "geekmaster" wrote:
    The developers must have ignored the naysayers who claim that positional motion capture with IMUs is impossible (or at least not good enough for head tracking).


    Well generally I think the naysayers mean that position for a single sensor cannot be determined. Kinematic constraints offers a neat little "cheat" to this limitation but still requires a number of sensors to work well.

    I am happy to see people working on this though - bucking the conventional wisdom about optical mo-cap. IMU mo-cap could have major benefits in terms of the variety and scale of environments.



    And you can squeeze a full set moCap system into a lock-lock box.
    More important... it's seriously much cheaper...


    my korean brand lunch box... :D
    http://zecken-und-dornen.de/lock-lock-dose-rechteckig.html
    rldaiAuthor
    Up and Coming
    June 24, 2013
    "blazespinnaker" wrote:
    Very cool. Do you have any examples with two or more people?

    Also, do you use it in conjunction with a camera as well?


    not yet for multi-char capture but I'm working on it everyday now.

    and yes we are also working on a syncronized camera recording too :D
    geekmaster
    Up and Coming
    June 24, 2013
    "rldai" wrote:
    my korean brand lunch box... :D
    http://zecken-und-dornen.de/lock-lock-dose-rechteckig.html

    Lock & Lock Dose, Ein Must-Have für jeden Cache-Owner

    Ah, I see... ;)
    This work is licensed under a Creative Commons Attribution-ShareAlike 3.0 Unported License.
    rldaiAuthor
    Up and Coming
    July 13, 2013
    hey guys, some recent update on VR with motion capture to share here
    It's in Chinese but I believe you'll understand what I mean
    3 screens from left to right
    1, from your goggle, 2. 3rd party view, 3. real-time captured video
    http://youtu.be/EGZ5Y5x95P4

    and... we are flying to Anaheim, California for Siggragh 2013 exhibition next weekend and the week following. Come and check us out!
    http://youtu.be/R6pnPggN0Tk
    Show more replies
    Terms & ConditionsAccessibility statement