PDA

View Full Version : Ok, the octo, but FPV!



Honk
08-03-2011, 08:02 PM
http://www.youtube.com/watch?v=fADI5XxAon0

So, the FPV cam mounted on the camera mount, but only gets servo'ed around roll axis. Was easiest that way. Worked ok, but at some point there because either interference due to bad angles or it may have been the low sun getting the FPV cam to flip out and I had to turn around... I thought I was gonna be able to sit in my chair for a while and feel a little chairborn.

As others say, it's hard to hover in FPV mode, especially with a massive craft in a tiny garden with trees and a house. But I think it's manageable. Also it's been like half a year since last times I tried FPV:ing.

I'm quite satisfied with the craft now, still experimenting with the PID settings, the motor mixer settings and the mechanical sensor arrangement to tune in the mechanical low pass filter. It's definitely tunable, but a hell of a lot harder to tune than a quad.

EQ1
08-03-2011, 08:48 PM
Ouch, Honk, it is wobbling more than ever! What did you do? It was quite nice in your last video, but now ...

Another question: Your camera is the same that I have on my first quad. But that gave me extremely low quality, especially when in motion, i.e. fast changing pictures. What are you using as equipment?

Honk
08-03-2011, 10:59 PM
The first video was much more wobbly, wasn't it? Now I think it happens more seldom. I did experiment some with the tips from Mikro, to increase gyro P to 160 and gyro D to -600. That worked kinda ok, but then I thought the pitch axis was wobbling more than the roll axis, so I decreased gains for the pitch axis, but maybe I'm deluding myself because of that roll relationship with the front/back motors you explained, expressed as pitch wobble. So maybe I need to have lower roll gains than pitch gains?

Also the flight before this one that might have turned out better, it was calmer winds. It's still pretty sensitive to wind. So I need to keep tuning and maybe add more stiffness to the rubber mounts in pitch axis only since that's where the most wobbles come from I believe.

Well, that camera is some cheap chinese non-brand camera, I'm sure it has many names. The system is working on 1.1GHz and the receiver says "Anboqui". Only. When I first did FPV, I did use my small laptop and a USB grabber inbetween, but that was really slow and causing latencies. I somehow managed to fly still... Now in this one I used a super cheapo B/W TV screen from the time humans used pen & paper (100% analog). Gets extremely fast response in video stream, so the camera/receiver have no problems with getting the flowing stream going well. In 30fps I believe. Those yanks...

EQ1
08-03-2011, 11:20 PM
Did you try to mix the motors according to their distance? Every P or whatsoever will affect EVERY one of them, so no solution for the physical problem from control theory here.

EQ1
08-03-2011, 11:22 PM
Well, that camera is some cheap chinese non-brand camera, I'm sure it has many names. The system is working on 1.1GHz and the receiver says "Anboqui". Only. When I first did FPV, I did use my small laptop and a USB grabber inbetween, but that was really slow and causing latencies. I somehow managed to fly still... Now in this one I used a super cheapo B/W TV screen from the time humans used pen & paper (100% analog). Gets extremely fast response in video stream, so the camera/receiver have no problems with getting the flowing stream going well. In 30fps I believe. Those yanks...

Yes, something like that China ones. I had a 2.4 Ghz receiver and transmitter but the quality was WAY behind that you posted.

Honk
08-04-2011, 12:11 AM
Well, I have calculated what the proportion between them distance wise should be, but meeh, this task does not at all sound like one that needs precise numbers so I figure going from 75-80% on the end motors going gradually up to 100% for the other end with equal steps should do it, as it is now. In these tests however I'm only running on 7 props all the time because of one of the batteries are too weak, which is why I'm trying to get the new bought one to work so I could try with all 8 motors again. I bet that makes some difference too. When the 8th motor go out in the beginning on the flight I immediately get a roll/yaw response and then it's back to the usual wobbling.

Well, I have a hard time wrapping my brain around what frequencies are the most suited ones for the task... Have not been able to experiment enough with video transmitting at all.

EQ1
08-04-2011, 08:25 AM
Well, because besides the motors, which are by far the heaviest parts, there are only a few grams from the arms. So the point mass approach will be quite accurate. Then it is essential to respect the fact that the dependency is hyperbolic. You are using a linear change for them. I would try doing it according to the model. Takes not much effort though.