Oh, the things you can do with a Vicon motion capture system and an unlimited research budget! (More coolness from ETH Zurich, via Robots.net)
nice work. but i really wonder what is the real life application of those systems that need external cameras? sure it drives innovation sooner or later in related real life applications, but i can't think of any right now.
do they at least release some code?
I'm thinking the poor intern who charges the batteries needs his name in the credits
The key is to reverse the situation. Put the cameras and processing on board and look for cues in the environment for spatial orientation.
ETH is working on this too. I finally have all of the parts necessary to assemble one of their Pixhawks.
why is nobody watching from the windows !
@arashi- Regarding turning a Vicon system inside-out: We already did this three years ago... http://www.diydrones.com/profiles/blogs/micro-helicopter-hovering-in Actually I got the idea of turning a Vicon system inside-out after seeing a talk by MIT Prof. Jonathan How, the first person (with many smart people in his lab) to use Vicon systems in this manner.
The above video is still impressive! Perhaps someday we can have little decorative drones flying around indoor atriums- they would be like birds except they don't, ummm, deposit stuff randomly.
@Mike : I thought they had automatic landing to a recharging pad ?
yes, i have seen it, i don't know if it is always used. a very different design from what I have been planning, and I have been thinking a lot about it. but i think i will maybe still go with my original design, but I was glad to see it and to consider it.
but I can imagine the intern all the same, always managing batteries, like a modern day cinderella...
One reason why reversing the vision system paradigm can't work (today) is CPU and battery power.
Vicon systems are sampling in the Khz range, were we all know most quadcopter flight boards sample (for the IMU) at 12-22Mhz (though the Arduino Due will greatly change that at 92Mhz :) ). And the resolution of the vicon imagers are in the megapixel range (at least the more recent ones). Put all that capability on a multicopter and your weight and battery limitations goes out door (1min flight time, 10lbs?)
Those that have tried to minimize the payload, VGA at 30fps (x4 for that matter), won't cut it for the maneuvers ETH is achieving. Mind that, now make it all work outdoors (!). Makes for a very, very challenging problem.
Now there are a number of folks trying to use sonar, Kinects (still heavy for outdoors), and lidar (Stanford), which look more robust, but processing power isn't there yet. The closest I've seen is with the AR.Drone: it has the cpu power (400+mhz DaVinci)--but the vga cameras run at 15fps, battery life is quick, and the motors have no real lifting capability. Then again, it can achieve 2-3 levels down from the ETH system: http://www.youtube.com/watch?v=X4C7iZZzyBM&feature=related
@Mike, yes the battery person needs credits. Lots of LiPos.
does it need to be megapixel? i remember that bee vision auto pilot posted a few days ago and they work with quite low resolution and recognize patterns.
Join DIY Drones
Welcome toDIY Drones
Sign Upor Sign In
Or sign in with:
© 2020 Created by Chris Anderson.
Report an Issue |
Terms of Service
Please check your browser settings or contact your system administrator.