Closed GoogleCodeExporter closed 8 years ago
Bambum0125 is Johann Van Niekerk .
Original comment by bambum0...@gmail.com
on 1 Jan 2011 at 4:15
Thanx to Antony Bee the math should be simple and here example
http://www.diydrones.com/forum/topics/regarding-apm-a-servo-question?commentId=7
05844%3AComment%3A245252&xg_source=activity
Original comment by bambum0...@gmail.com
on 1 Jan 2011 at 10:34
I take it "P/T" means pan-tilt camera control? If so, this functionality has
been discussed in the dev team (I think Jason was planning on trying to
implement it in the beginning of December).
However, I don't think this is nearly as easy as you think, even assuming
perfect camera mounting where the pan axis is the plane's vertical axis when
level and the tilt axis is contained in the plane normal to the pan axis. To
see why, try to figure out what pan and tilt angles are required for a plane
oriented with 10 degrees pitch and 10 degrees roll to point at a target that is
20 meters left of it, 2 meters in front of it, and 10 meters below it.
Even if the problem is simplified to assume that the target is always directly
off the plane's wing (thus negating the need for the camera's pan axis), extra
control code would have to be written to track the yaw exactly, and it would
only work when the pitch was zero.
It is certainly possible to construct an algorithm that compensates for all
these factors (and even allows camera rotation schemes other than pan-tilt),
but I'm not currently prepared to release my algorithm.
Original comment by bjpcalt...@gmail.com
on 6 Jan 2011 at 11:17
Yes PT means Pan Tilt System , I understand for 100% accurate pointing alot of
variables has to be concedered and worked out probably Gyro on camera would
help , but what I was thinking of was simple code that would just turn pan left
or right by a # defined degree / relevant to loiter radius and tilt calculated
by altitude and loiter radius to center .?
I know it won't be accurate but with a 45* field of view on a these Sony Had
CCD"s it won't be that big of
a problem, I'd say only when big zoom functions comes in accuracy would be
needed to look because field of view would change .
So I would think that what the P/T system would need to be told by APM is
When entering loiter clockwise set pan 45degree right when antclockwize turn
pan 45 degree left then using TAN. Work out the angle needed for tilt because
hight you have and distance you have ( loiter radius) so angle could be
determined .
Note: this isn't really for FPV viewing but more for mission planning before
flight and something extra for people with outputs left open on there APM's
Original comment by bambum0...@gmail.com
on 7 Jan 2011 at 12:12
How about just stabilization at first, with neato stuff like manual point and
auto track later... should be easy based on IMU data alone... aircraft
banks/turns, pan and tilt to compensate...
Original comment by Draconis...@gmail.com
on 8 Jan 2011 at 4:35
That is a Idea but wouldn't the Pan tilt system then be mounted on the same
level as the APM ? other wize IMU vaules wont help.
Original comment by bambum0...@gmail.com
on 8 Jan 2011 at 5:21
well, yes depending on location of the camera PT mount/gimbal... programmed
offset or manual nudging should be enough dont u think?
Original comment by Draconis...@gmail.com
on 8 Jan 2011 at 5:25
True yes. But I think the fether away from the APM the more inaccurate when
aircraft rolls it should work great but what if it pitches ? The PT system
going to be somewhat 20cm above the APM .... I think that might pose a problem.
Or how would you go bout it?
Original comment by bambum0...@gmail.com
on 8 Jan 2011 at 5:48
if based on the loiter mode for example, then distance from target should be
limited to + - what the platform / PID will keep it going round. just trying to
suggest starting with the simplistic approach of linking the P/T to the gyros,
as you would in a simple stabilized gimbal setup.
But as you stated earlier, it would most likely depend on keeping the camera
near the cg. If the camera is on top, then it should be able to see 180pan with
90tilt to see from wing level to horizon+ in a bank. if its hung below then
typicaly the wing cuts off visibility to alot of ground in a bank.
this is why most have camera forward of cg. or better yet in a clear pod out
front.
Original comment by Draconis...@gmail.com
on 8 Jan 2011 at 9:13
Yes that makes sence to me as well and I plan to install my camera close to the
CG , would it be a option to add a little gyro board to the pan head and let
APM handle it from there or are we better of getting a dother board to handel
that.?
below is a link t the origenal Arducam program that triggered camera when
arriving at way point. see files attached.
With alot of help from Ritchie it should be easy just to add this onto that code
void CMD_PAN(int angle){
if angle < -45 {
angle = -45;
}
if angle >45 {
angle = 45;
}
servo_out[CH_5] = angle;
}
this should allow the first part of this enhancment to be complete but as you
said stabilization is needed to ,what would be your thoughts?
Original comment by bambum0...@gmail.com
on 9 Jan 2011 at 10:01
Attachments:
Original thread here
http://diydrones.com/forum/topics/photos-with-apm-issues-tweaks?xg_source=activi
ty&id=705844%3ATopic%3A213509&page=2#comments
Original comment by bambum0...@gmail.com
on 9 Jan 2011 at 10:02
check out this page . it already covers tracking for objects , implemented with
Arduino , even the code is present .
http://umassgv.blogspot.com/2010/07/realtime-tracking-with-pan-tilt-camera.html
Original comment by abx...@gmail.com
on 9 Jan 2011 at 2:55
Abx666 you made my day.....
I saw his video on youtube but could nt get his link to wotk thanx. have you
built it?
Original comment by bambum0...@gmail.com
on 9 Jan 2011 at 3:10
Thanx Abx666
Here is target tracking code written in arduino
can we implement it?
Original comment by bambum0...@gmail.com
on 9 Jan 2011 at 3:22
Attachments:
no but im planning on .. i need to know wheter it would be smarter to
integrate into the I2c or the regular OUTS
Original comment by abx...@gmail.com
on 9 Jan 2011 at 3:42
I think we will soon find out ...I put a link on front page in"pan tilt blog"to
get people to look at it.
Original comment by bambum0...@gmail.com
on 9 Jan 2011 at 3:45
id be glad to help in any way.
Original comment by abx...@gmail.com
on 9 Jan 2011 at 4:07
One question . what will we be tracking from a Airplane?
Original comment by bambum0...@gmail.com
on 9 Jan 2011 at 4:25
cars ? people ? geo features?
Original comment by abx...@gmail.com
on 9 Jan 2011 at 5:53
Hehe abx666
I like the Geo features idea ;-)
How is your programming skills?
Original comment by bambum0...@gmail.com
on 9 Jan 2011 at 6:18
A team member (Joe Holdsworth) has been added to the dev team to implement
camera tracking functionality.
Original comment by dewei...@gmail.com
on 10 Jan 2011 at 12:23
Awesome thanx .
Looking forward to it!
Original comment by bambum0...@gmail.com
on 10 Jan 2011 at 12:28
prototype version in operation here:
http://diydrones.com/profiles/blogs/camera-gps-targeting-assisted
Original comment by joeholds...@gmail.com
on 10 Jan 2011 at 1:22
[deleted comment]
yea i can help out with code , but i need to first get all the hardware
required , which might take a while .
Original comment by abx...@gmail.com
on 10 Jan 2011 at 4:22
Hi Joe Great work !
I added this one without seing your blog about it, must have been before i
joined but cant wait!
Thanx for sharing your hard work
Original comment by bambum0...@gmail.com
on 10 Jan 2011 at 6:21
Most of the points here are already implemented in the APM_Camera branch. And
are documented on the wiki. But some more tests are needed.
Original comment by amilcar....@gmail.com
on 12 Apr 2012 at 11:24
The new AP_Mount code now supports this.
Original comment by tridg...@gmail.com
on 19 Jul 2012 at 5:28
Original issue reported on code.google.com by
bambum0...@gmail.com
on 1 Jan 2011 at 4:07