I know that Arducopter2 supports camera stabilisation, so what I would like is to have an APM that is dedicated to just camera stablisation.
Can anyone help by cutting down a version of the AC2 code to just have the APM board do camera gimbal stablisation.
I would like to use this is a standalone R/C camera gimbal stablisation system without all the multicopter code.
I know I can just buy a FY30A stabiliser but, it does a crap job of camera gimbal stabilisation.
Replies
What happened to this thread? I was finding it interesting....
Hi Hai,
It should be pretty simple, just keep the radio.pde file and the AHRS, RC_Channel, and AP_Mount classes.
Throw the rest away, and there you have it.
Regards,
Amilcar Lucas
The camera stabilisation function has to change. And I am working on it. The way that a professional system works is by utilizing the the inner and outer axis principle. All of the camera imu's that I have ever seen for models are basically an AHRS that tries to fly the camera base plate and keep it level. That is a no brainer because what you have done is simply create a second outer axis that can only make a small improvement at best. Your model is the outer axis. When your imu is placed on the camera mount it will not see the very small initial movement because the mount is suspended (typically silicon rubber mounts) so the mount is already behind the disturbance. The next problem is the latency within the sensor output. If you remember how a typical tail rotor gyo oscillates when you try to crank up the gain, you can see that any feedback loop will not be accurate enough. The MPU 6000 has variable sensitivity settings so we need the 250Deg/sec and the +_2 G I am reliably informed that this could result in a stabilised pointing accuracy of .005 deg. This is pretty close to the Cineflex .003. Such sensitivity would be able to see you heartbeat as you touch it.
WOW what a selling point.
So what we need is that inner axis stability which can only work with the imu attached to the model correctly. It will measure the actual disturbance of thr model and create a mirror image of that which is then fed to the servos. No feed back loop, no sensor fusion alg. just a nice simple accurate solution. Also goodbye to large floppy high inertia camera mounts DIY drone will eventually have a tiny direct drive knuckle joint that will be dedicated to the system.This what is needed for these small cameras which are set to get even smaller.
The mechanical resolution and how fast the servo can move makes one sort of difference as to how it will move smooth.
Pushing on a servo will make a pulse these pulses can also make vibration to a camera.