CSU Chico Tilt Rotor Flying Machine, (Airframe and Electronics, 2 of 3). / General Science and Electronics / Forums

Forums

4hv.org :: Forums :: General Science and Electronics

« Previous topic | Next topic »

CSU Chico Tilt Rotor Flying Machine, (Airframe and Electronics, 2 of 3).

first previous 5 6 7 8 next

Move Thread

LAN_403

AndrewM

Tue Jul 10 2012, 05:44PM

Registered Member #49 Joined: Thu Feb 09 2006, 04:05AM
Location: Bigass Pile of Penguins
Posts: 362

That code appears to be differentiating position feedback to get velocity. You, on the other hand, have a gyro so you can get velocity data directly.

Patrick

Thu Jul 12 2012, 02:34AM

Registered Member #2431 Joined: Tue Oct 13 2009, 09:47PM
Location: Chico, CA. USA
Posts: 5639

IMPROVED!

its now easy enough for any dope to fly.

AndrewM

Thu Jul 12 2012, 07:04AM

Registered Member #49 Joined: Thu Feb 09 2006, 04:05AM
Location: Bigass Pile of Penguins
Posts: 362

Wee, looking better.

Are you still limited to hops like that or have you achieved longer flights?

Patrick

Thu Jul 12 2012, 07:07AM

Registered Member #2431 Joined: Tue Oct 13 2009, 09:47PM
Location: Chico, CA. USA
Posts: 5639

i can muddle along in 10-15 second sloppy hovers.
If i change my dual rates and exponential i think i can get it stable for minutes at a time. Ill do this first thing tomarrow.

Patrick

Sun Jul 15 2012, 02:58AM

Registered Member #2431 Joined: Tue Oct 13 2009, 09:47PM
Location: Chico, CA. USA
Posts: 5639

ok Ive been doing some thinking...

i want to use a ramp, to initially change acceleration states, this will avoid wind-up errors in domain, while setting a maximum velocity allowable will avoid wind-up in range.

ive also been looking at tinkerforge:

Patrick

Tue Jul 17 2012, 05:29AM

Registered Member #2431 Joined: Tue Oct 13 2009, 09:47PM
Location: Chico, CA. USA
Posts: 5639

new vids

the first step, a PI loop.
SP = setpoint (input form pitch stick), m/s
PV = process variable (output for feedback), m/s
MV = Manipulated variable (control signal out), degrees of tilt.

so for this PI loop , i would assign meter per second values to pitch stick position (fore and aft speed) then use the accelerometer to judge the current velocity, and adjust as needed. Then repeat as funtion prototype calls for each axis.

but as for the ugly part, the gyros must be able to contribute too !? so how do i let the accelerometers have there say and then the gyros? Obviously, the accels are responsible for translation, the gyros for rotation.

is it possible that i have to run two PIDs in parallel one the the gyro's commanded tilt (position PID), the other for the accelerations needed duration, (the velocity PID?).?

EDIT: if the gyro PI tilts then degees/seconds are output, while the Accel PI keeps track of velcity/time. If that is the case then i think the two must be in series, not parallel as i wrongly stated above..

WaveRider

Tue Jul 17 2012, 01:44PM

Registered Member #29 Joined: Fri Feb 03 2006, 09:00AM
Location: Hasselt, Belgium
Posts: 500

If you want full sensor fusion between accelerometers , gyros, even visual, GPS and user data, you can't beat the Kalman Filter approach. You are on the right track with the PID controllers. What is missing (as you discovered) is a way to put all the sensor data together. The Kalman filter's ability to handle vectorized data is very convenient..but can get cumbersome particularly if non-linear physical models are used.

Patrick

Tue Jul 17 2012, 08:24PM

Registered Member #2431 Joined: Tue Oct 13 2009, 09:47PM
Location: Chico, CA. USA
Posts: 5639

OK, if started a post with DIY Drones, under the screen name "Test Pilot Mafia" unfortunately my post is going unseen, this has happened before in other places too.

Me ->

edit: pendulum-cart model :

bicopter:

Quote: "The SCORPID-500 is now fully stabilised and controlled by a 9 DOF IMU (ArduIMU+ v2, HMC5843, GPS with firmware TriStab v3.3 JLN). The SCORPID-500 VTOL UAV uses an innovative design based on Gary Gress concept from Gress Aero. The Oblique Active Tilting (OAT) at 45Â° of the twin engines allows a full pitch control by using the induced gyroscopic moment."

OAT = oblique active tilting,

note the angle allows a single tilt axis to cross both the roll and pitch planes!

Patrick

Wed Jul 18 2012, 09:33PM

Registered Member #2431 Joined: Tue Oct 13 2009, 09:47PM
Location: Chico, CA. USA
Posts: 5639

OAT = Oblique Active Tiliting, singular or dual.

the OAT allows landing on tilted surfaces, which quads and tri's cant do.

PDF:
]oat.pdf[/file]

Ive been told/fear that possibly my current design wont work. supposedly (and theres no documentation from kkmulticopter) you would need a low (long vertical destance) CG and a very low inertia prop/motor combo. or the system has no hope of becoming stable.

My design was the typical FAAT type tilt rotor, but the OAT may be far better, especially for smal compact UAVs.

Patrick

Tue Jul 24 2012, 02:32AM

Registered Member #2431 Joined: Tue Oct 13 2009, 09:47PM
Location: Chico, CA. USA
Posts: 5639

Its still wont hold a hover but at least i can show definate pitch control.

OK, with the above ive just demonstrated that :

1) yaw works as predicted.
2) using gyro scopic forces does indeed effectively control pitch.
3) the gyro prop idea (Greg Gress) seems to still work with the KKmulticopter/flycam/twin copter board.

So, OAT is defineatly superior to FAAT (fore aft active tilting)...
I plan to switch from the primative KK/flycam baord to the HK 2.0 board seen here.

the problem with the flycam boards is that they seem primative, with no advice or specs given to build the airframe, for 140$ you get the board and some simple instructions and settings, but nothing more. And ive been told that the KK boards dont have PIDs, Kalman filters or any of the usuall stability strategies, if true i dont know how it works, just those three magic potentiometers. it has only gyros.

the HK 2.0 board has an LCd screen, buttons, and an editor to change PI, PD, PID type settings, for each axis for upto eight motors, it has both accelerometers and gyros. All for just $34.

first previous 5 6 7 8 next

Moderator(s): Chris Russell, Noelle, Alex, Tesladownunder, Dave Marshall, Dave Billington, Bjørn, Steve Conner, Wolfram, Kizmo, Mads Barnkob