I've been reviewing the latest developments in "Adaptive Control" as opposed to PID control and came across this video of a quad adapting to a broken rotor blade. Hopefully, we will soon use these types of algorithms in our own diydrones projects.
Adaptive, robust, LQR, LQG etc are very nice/powerful control techniques. But if you read carefully the .pdf attached here (or you can look them up in the wikipedia), you have to know all the system matrix, and this is by far, the most difficult task, you have to find out a reliable model of your particular system.
And of course, we are assuming a linear system (closer to the equilibrium like in this video), if you aims for a not linear system, the algorithms grow in complexity.
So, for control basics, PID is by far the simplest way, without knowing a super detailed information about the system. And it will works without any "super-disturb".
Adaptive, and other techniques, if you don't have a detailed mathematical model of the system, could be less reliable than a simple PID controller.
BrainyBEE Autopilot
Aerojeep's BrainyBEE autopilot is the world's smallest, yet the most versatile single DSP flight
controller. The powerful DSP executes a highly optimized multi-state extended Kalman Filter to
fuse INS sensor inputs with GPS positioning data; Sophisticated guidance includes up to 100
waypoints of various types; Advanced neural network adaptive control technology delivers robust
tracking performance throughout the full flight envelope. The navigation, guidance and control loops
run at 50Hz to ensure high performance flight control.
X4-flyer Trajectory Tracking (Neural Adaptive Control )
In following video, X4-flyer (quadrocopter) performs fully autonomous trajectory tracking, equipped with Aerojeep's BrainyBEE autopilot, which is the world's smallest high performance autopilot, embedded with neural network adaptive control technology, and capable of providing integraded guidance and control of a UAV network of different types including fixed wing, helicopter, X4-flyer, airship and other VTOL UAVs.
I don't really see how Adaptive Control is different than making the weights used in PID control variable/based on the current conditions of the craft. Ie, assign each "rotor" a "lift" value. If, at the maximum correction situation, the craft does not correct than adjust the lift values of the rotors. Am I missing something?
Hi Greg ! (Prof. GK Egan, the great Quadrocopter Guru), There is PID type of contol method in the firmware of your UAVP, UAVX & UAXNav, I think adaptive type of control method is much better and would highly be desirable to be implimented in your Quadrocopter firmware.
It is evident from the video that adaptive control is far far better as compared with PID control.
My special thanks to you for bringing this thing in the eye of DIY comminity !
Aside to Billu Bhaiya & Ben Levitt, Jordi: It is certainly a very nice thing and can be adopted in UDB & Ardupilot.
Welcome! Future Home of Another Amazing Website Powered By Exabytes
Comments
BTW does anyone know what price range this BrainyBEE AP sells for?
Adaptive, robust, LQR, LQG etc are very nice/powerful control techniques. But if you read carefully the .pdf attached here (or you can look them up in the wikipedia), you have to know all the system matrix, and this is by far, the most difficult task, you have to find out a reliable model of your particular system.
And of course, we are assuming a linear system (closer to the equilibrium like in this video), if you aims for a not linear system, the algorithms grow in complexity.
So, for control basics, PID is by far the simplest way, without knowing a super detailed information about the system. And it will works without any "super-disturb".
Adaptive, and other techniques, if you don't have a detailed mathematical model of the system, could be less reliable than a simple PID controller.
Aerojeep's BrainyBEE autopilot is the world's smallest, yet the most versatile single DSP flight
controller. The powerful DSP executes a highly optimized multi-state extended Kalman Filter to
fuse INS sensor inputs with GPS positioning data; Sophisticated guidance includes up to 100
waypoints of various types; Advanced neural network adaptive control technology delivers robust
tracking performance throughout the full flight envelope. The navigation, guidance and control loops
run at 50Hz to ensure high performance flight control.
In following video, X4-flyer (quadrocopter) performs fully autonomous trajectory tracking, equipped with Aerojeep's BrainyBEE autopilot, which is the world's smallest high performance autopilot, embedded with neural network adaptive control technology, and capable of providing integraded guidance and control of a UAV network of different types including fixed wing, helicopter, X4-flyer, airship and other VTOL UAVs.
Extract from (Predictor-Based Model Reference Adaptive Control - Paper)
My special thanks to you for bringing this thing in the eye of DIY comminity !
Aside to Billu Bhaiya & Ben Levitt, Jordi: It is certainly a very nice thing and can be adopted in UDB & Ardupilot.
-
1
-
2
-
3
-
4
of 4 Next