Heres the idea:
Objective: making a quad that can loose a motor or esc without it crashing.
Solution: Mount four motors under the current four motors. Mount props that turn the same way. So there is still yaw control. So you still have 4 turning cw and 4 turning ccw.
Then connect the signal cable from both motors on each arm to each other and then to the APM.
If my theory is correct, if you had to lose 1 motor, the one underneath it or above it would compensate for the loss. It probably wont fly well, but should not crash and burn...
I dont know if any code needs to be changed? Im going with no or maybe a little.
What say the experts?
:)
Replies
In this case the controller puts less expectation on the failed actuator, thus the duty of the failed actuator will be less and more expectation goes over the healthy actuator right over it.There are many techniques to do so such as Pseudo inverse, Fixed point,but the overal control technique is Control Re-allocation. But you should keep the baseline controller which in this case is a cascade PID. Here is my video doing exact same thing :
https://www.youtube.com/watch?v=Kj3_pvKWiNk
Good luck
I want to try this as well
There will still be only 4 channels going to the ESCs
Each channel will be split into two ESCs, bottom and top ESCs will have different LiPo pack.
This way, unless one channel goes completely out, it should be redundant enough to stay more or less level (assuming the motors are powerful enough) - it will survive LiPo going dry and motor or ESC failure.
There needs to be 8 motors and 8 escs. All in pairs.
Using a quad makes it much easier for installation, its all hardware very little software.
From my limited knowledge, i think it would work fine. When I have more motor and escs, im going to try it!
To test how well it flies with a faulty motor, i just unplug one...
I'm not your expert, but I've seen a few configurations, and know a touch of electronics. In the stock motor/ESC combination, remember in your design that the ESC is often paired, in terms of Amperage such that it can safely drive the motor at full throttle and then some (wind forces on the prop increase work/power requirements to turn it) Pull too much current for an ESC, say by attaching it somehow to two motors...
The coaxial approach to motor mounting, in general is often done, in Y6, X8, for example. I have read and heard reports that it is as much as a 30% "loss" in thrust over fully distributed motors, and I would not be surprised if it is higher (or lower) as I haven't seen the math.
Driving several motors from a single channel, in general, has been demonstrated recently in posts here. The major examples I recall are the various Vtail8 blogs and video and video, but bear in mind that this is a very different motor distribution. I think this is sometimes done in other V-configurations and other non-squared polygonal/parallel configurations. And I believe each motor has a dedicated ESC, only the speed control signal/output channel is combined. Some objections were raised, but he clearly flys all the same, whatever detractors or benefits there may be.
But fault tolerance, it is my understanding, is a feature of a well-built octo in a more traditional design, and I suspect a traditional (i.e., no unusual wiring) coaxial octo would likely tolerate one or even two motor failures under sedate conditions.
I have only purely theoretical knowledge to go on, but hope you find some of the videos interesting. All my frames are quads so far, but I got a shipment of ESCs and motors in this week for an octal design. But I need to run more simulations and calculations before I try it, and nothing so experimental for me for my first 8 build. Too much money in the air ;) I'm all custom frame and wacky ideas with a quad, but I get squeamish over an octo (comes with having mostly mid range castle ESCs in inventory at $80 each *wink*)
If you haven't read this page already, you'll want to: http://code.google.com/p/arducopter/wiki/AC2_Multi