Hi all,

I would like to make a project in which I would like to create a software that makes the aircraft fly level when the pilot loose orientation? The module will be triggered once the pilot returns the sticks to neutral.

Attached you can find a simple diagram that describes what I have in mind. It shows that I would like to pass the control of the right stick (in mode 2 TX) to the Level Control Box which will not make any changes on the signals given from the receiver, when when the stick is on neutral position it will be triggered and make the necessary adjustment on the Aileron and the Elevator to make the model fly level.

Regarding the Sensor unit I intend to install two small cameras sensors (one looking forward and another to one of the sides of the aircraft) that look at the horizon and through a software function will return the roll / pitch angles of the aircraft.

What is your feedback about this issue?
Will it be easy to read the signals that are produced by the receiver and fed to the servos?

Hi,

Developing a system like this should be fun and not too difficult, with the exception of horizon sensing itself. That is always tricky. If it is your objective to take on the challenge of finding the horizon reliably using cameras then let the games begin! But if you don't mind avoiding the sensor development so that your project can move on towards your larger goals, I suggest that you check out the FMA Copilot system. It works remarkably well, is inexpensive, and the output signals can be easily adapted for what you have in mind.

Good luck!
Q

thanks Quaker ..... but my intentions is to replicate the Co-Pilot system myself :-)
The problems that I am seeing are the interfaces Camera-CPU, CPU-Servo Controller and also the vibration caused by the engine that my disturb the camera.
See this vid that I took: http://www.halfarmodelflying.com/art...&title=Gallery

The servo control and stabilisation are pretty straight forward.
A good place to start would be having a look at the open source code of Ardupilot on DIY Drones in their version 2 software.

They are using the FMA sensors, as is nearly everyone else producing a cheap autopilot, so you might be onto something if you can do image analysis instead of using the IR sensors.
I would see that as the biggest challenge.
Although a camera in each direction would still cost a lot more than the FMA Thermopiles.

how will you deal with the aircraft attitude at 90 degrees to the horizon, or inverted? From a visual perspective, how can the software tell the difference between the land and the air? that seems to be the biggest challenge.

hi,

the issue would be only when inverted.
This project was discarded due to hardware prroblems..... it is very difficult to implement the required software on a simple IC.

http://www.roborealm.com/ might be of some interest in this area