@nerdcaster how about the fact that computers use inanimate objects to make '' cool stuff'', lika metal maze for example, while the brain has complete control over human bodies which are capable of making stuff that makes''cool stuff'', like a factory or building!!!!!!!
I was hoping to use the CMUcam3 (352x288 at 26fps) camera along with a microcontroller (maybe a Freescale MCF5213 series) that could do DMA transfers, has multiply-accumulate capability, preferably sufficient RAM to hold an image with bus interface speed that is at minimum equal to the pixel data rate. I would still need video post-processing display as well. I like Analog Devices BF-561 surveillance kit but cost is the main problem. I agree that Labview is the tool of choice.
I expanded your description so now I realize your team used LabView. Was all position and velocity data extracted from the vision system? How was table tilt sensing implemented? What were the original system specifications? What are the camera specifications?
Yes, position is extracted from the vision system (key function: IMAQ detect circle and edge detect). Velocity is then calculated by simple discrete differentiation. Table tilt angle is not measured, however it is estimated from motor position (step motors - 1Nm nom. torque).
Camera: Sony XC-HR50 Grey Scale, 60fps, 648x494 pix
Not sure what you mean by original system specifications, we build it from scratch :-)
2/2 I think the biggest challenge by implementing in a microcontroller is the image processing, which is greatly simplified by using the IMAQ tools in Labview, but a microprocessor system could be very nice (and real time!).
Did your team use LabView for the vision GUI and mapping? I assume you also use Labview for motor control and any control laws? Did your team publish your results anywhere? I'd like to do the same project but maybe I'll see if I can embed the vision processing and control laws in a microcontroller? How big are your motors? Are your motors steppers? That is a sweet implementation and great work gentlemen!
That is freaking fabulious!
FlyMario2 3 months ago
Sweeeet!
AsbjornOlling 4 months ago
Very very nice control system you've got there!
pirsquared99 4 months ago 3
yo, human brain, its computer, how do you plan to one up me this time??????
nerdcaster 4 months ago 5
@nerdcaster how about the fact that computers use inanimate objects to make '' cool stuff'', lika metal maze for example, while the brain has complete control over human bodies which are capable of making stuff that makes''cool stuff'', like a factory or building!!!!!!!
mpaskey1 3 months ago
@mpaskey1 you win this round
nerdcaster 3 months ago
lækkert lækkert!
Tripleyou 4 months ago
Wow. That was simply amazing to watch.
Well done, guys!
fishesgotfeelingsto 7 months ago
YOU CAN WATCH LOT OF HUMANOID ROBOTS ON MY CHANNEL
luisbeck007 7 months ago
Nice!
specallez 1 year ago
I was hoping to use the CMUcam3 (352x288 at 26fps) camera along with a microcontroller (maybe a Freescale MCF5213 series) that could do DMA transfers, has multiply-accumulate capability, preferably sufficient RAM to hold an image with bus interface speed that is at minimum equal to the pixel data rate. I would still need video post-processing display as well. I like Analog Devices BF-561 surveillance kit but cost is the main problem. I agree that Labview is the tool of choice.
krimoc 1 year ago
I expanded your description so now I realize your team used LabView. Was all position and velocity data extracted from the vision system? How was table tilt sensing implemented? What were the original system specifications? What are the camera specifications?
krimoc 1 year ago
@krimoc
1/2 Hey krimoc, thank you for your interest!
Yes, position is extracted from the vision system (key function: IMAQ detect circle and edge detect). Velocity is then calculated by simple discrete differentiation. Table tilt angle is not measured, however it is estimated from motor position (step motors - 1Nm nom. torque).
Camera: Sony XC-HR50 Grey Scale, 60fps, 648x494 pix
Not sure what you mean by original system specifications, we build it from scratch :-)
danromdk 1 year ago
2/2 I think the biggest challenge by implementing in a microcontroller is the image processing, which is greatly simplified by using the IMAQ tools in Labview, but a microprocessor system could be very nice (and real time!).
Thank you for the compliment!
danromdk 1 year ago
Did your team use LabView for the vision GUI and mapping? I assume you also use Labview for motor control and any control laws? Did your team publish your results anywhere? I'd like to do the same project but maybe I'll see if I can embed the vision processing and control laws in a microcontroller? How big are your motors? Are your motors steppers? That is a sweet implementation and great work gentlemen!
krimoc 1 year ago
Nice work! :D It's much faster than the DTU version! xD
01Maus 2 years ago