positional tracking with rotational only

Post Reply
LukePoga
Two Eyed Hopeful
Posts: 89
Joined: Mon Aug 20, 2012 3:49 am

positional tracking with rotational only

Post by LukePoga »

I was thinking of a potential way to solve positional tracking. could you make a tight fitting long sleve shirt and just stick a few dozen rotational sensors around the shirt to get perfect positional tracking. you can just work backwards with maths (virtual skeleton). there would be no drift as your ass/hips are the anchor point.

and you could put lots of points near the hand. even extend the shirt in to a glove. this gives perfect positioning of hand and torso with no new technology required.
geekmaster
Petrif-Eyed
Posts: 2708
Joined: Sat Sep 01, 2012 10:47 pm

Re: positional tracking with rotational only

Post by geekmaster »

LukePoga wrote:I was thinking of a potential way to solve positional tracking. could you make a tight fitting long sleve shirt and just stick a few dozen rotational sensors around the shirt to get perfect positional tracking. you can just work backwards with maths (virtual skeleton). there would be no drift as your ass/hips are the anchor point.

and you could put lots of points near the hand. even extend the shirt in to a glove. this gives perfect positioning of hand and torso with no new technology required.
This is the third thread discussing this idea, in which this document was linked:
http://www.cl.cam.ac.uk/research/dtg/ww ... thesis.pdf

It uses a foot as an anchor point, where it zeros out the velocity to get much more accurate positional data (about 0.3% accurate). I mentioned in those other threads that you should be able to use you seat as an anchor. And the seat is anchored to the Real World, so that small error could be adjusted out by detecting when you are sitting upright centered in your chair.
Krenzo
Binocular Vision CONFIRMED!
Posts: 265
Joined: Tue Sep 07, 2010 10:46 pm

Re: positional tracking with rotational only

Post by Krenzo »

There are products like this already: MVN and YEI

The problem is that your positions are all relative to your anchor (the hip in your case). The system will treat your hips as never actually moving, only staying in place. That's not very good for VR applications if you can't actually walk anywhere. That's ok if you're just doing motion capture for animation and can manually animate the anchor point later. If you put your anchor point on the feet, then you can approximate movement, but drift will still accumulate error in your absolute position.
LukePoga
Two Eyed Hopeful
Posts: 89
Joined: Mon Aug 20, 2012 3:49 am

Re: positional tracking with rotational only

Post by LukePoga »

thanks. i thought i was a genius for inventing it, but i guess not.
LukePoga
Two Eyed Hopeful
Posts: 89
Joined: Mon Aug 20, 2012 3:49 am

Re: positional tracking with rotational only

Post by LukePoga »

Krenzo wrote:There are products like this already: MVN and YEI

The problem is that your positions are all relative to your anchor (the hip in your case). The system will treat your hips as never actually moving, only staying in place. That's not very good for VR applications if you can't actually walk anywhere. That's ok if you're just doing motion capture for animation and can manually animate the anchor point later. If you put your anchor point on the feet, then you can approximate movement, but drift will still accumulate error in your absolute position.
yes correct. but all this doesnt matter with our wired oculus rift. i just want to use it in a seated position with games that i write. in the future when we are all running around wireless im sure people can use other technologies. but right now i just want something to use with my rift immediately.

so has anyone actually made one, besides in movies and military? i did read somewhere that it was 45 bucks for 100 sensor units? just number the sensors, map them to that location on virtual skeleton. its so easy to do what are you guys doing about it? where do you get those sensors i can do it myself lol.
geekmaster
Petrif-Eyed
Posts: 2708
Joined: Sat Sep 01, 2012 10:47 pm

Re: positional tracking with rotational only

Post by geekmaster »

LukePoga wrote:i did read somewhere that it was 45 bucks for 100 sensor units?
I am sure interested in those "two dollar sensors". Are you sure they are not just reflective optical targets, given that "too good to be true" price?
LukePoga wrote:where do you get those sensors i can do it myself lol.
The cheapest source I am aware of for IMU sensors is the STM32F3DISCOVERY development kit, for $10.80USD (including domestic shipping):
http://www.st.com/web/catalog/tools/FM1 ... 2/PF254044

Image

It was covered on Hackaday too:
http://hackaday.com/2012/09/12/stm32-f3 ... me-extras/

It is rather large compared to other dedicated IMU boards, but it is meant for prototype firmware development and brings ALL of the CPU pins out to sockets (similar to an Arduino shield), and it is very inexpensive compared to those other smaller boards. It also has a full ARM CPU on it (similar to the Raspberry Pi), so you can do more than just position sensing. You can build a much smaller unit that runs your firmware, by ordering individual components used on that dev kit PCB. If you want something you can order now that is smaller than that, you will have to pay a lot more.

I bought three of these boards, and I plan to integrate them into FreePIE and/or SDL, when I am ready to use them.

EDIT: I see that the price went up $0.10 since I bought them. It is now $10.90USD:
Digi-Key Price ($10.90): http://www.digikey.com/scripts/DkSearch ... 3DISCOVERY
Mouser Price ($16.54): http://www.mouser.com/ProductDetail/STM ... brug%3d%3d
Data Sheet: http://www.st.com/web/en/resource/techn ... 063389.pdf
User Manual: http://www.st.com/web/en/resource/techn ... 063382.pdf
Getting Started Manual: http://www.st.com/web/en/resource/techn ... 062662.pdf
And more: http://www.st.com/web/catalog/tools/FM1 ... r=70071840

Here is source code for a "gyro mouse" emulator that runs on the STM32F3DISCOVERY:
https://github.com/Laurenceb/STM32F3Discovery

Here is a video showing this board in use as a USB IMU (and showing its size relative to a human hand):
[youtube]http://www.youtube.com/watch?v=rqw5J4Ee6JE[/youtube]

EDIT2: Hacking the STM32F3DISCOVERY to unlock more features:
http://www.taylorkillian.com/2013/01/re ... -from.html
Last edited by geekmaster on Mon Apr 15, 2013 9:07 am, edited 3 times in total.
LukePoga
Two Eyed Hopeful
Posts: 89
Joined: Mon Aug 20, 2012 3:49 am

Re: positional tracking with rotational only

Post by LukePoga »

thanks for the info. I definitely underestimated the size of these units. I can see a single control board managing several sensor arrays on the limbs, but it would require substantial work and further miniaturisation.

Its not past a decent shop like Oculus tho. Fingers crossed something comes out because there are so many applications of hand control - its mind blowing.

even if its just a vest for head positioning thats ok too. i really wanna lean over a cliff to look down. none of this tapping the w key business.
User avatar
nateight
Sharp Eyed Eagle!
Posts: 404
Joined: Wed Feb 27, 2013 10:33 pm
Location: Youngstown, OH

Re: positional tracking with rotational only

Post by nateight »

geekmaster wrote:STM32F3DISCOVERY development kit, for $10.80USD (including domestic shipping)
What the what?! A fully broken out IMU shipped to my doorstep for about ten American dollars? :woot

All the struggling I've been doing on my secret project to cut the IMU hardware costs out of a dataglove suddenly seems kind of pointless. If this library I've settled on doesn't sit up straight and play nice in a big hurry I may have to reconsider the whole effort. :?
Shameless plug of the day - Read my witty comments on Reddit, in which I argue with the ignorant, over things that don't matter, for reasons I never fully understood!
Post Reply

Return to “Oculus VR”