If you want a -3dB corner frequency that isn't in the data sheet table:

1/2piRC

R in ohms, C in farads. Thus 1/2pi (110K ohm) (1 uF)

1/2pi (110 E3 ohms) (1 E-6 farads) = 1.45 Hz
Hardware_man

After reading the document named "Perching Using a Quadrotor with Onboard Sensing" of Jeremy C. Goldin from Utah State University (known as aggressive maneuvers with quadrotors) I decided to go in details with a PID controller... I think (as of now) Kalman is not the think I have to llok for.. I have to integrate a linear and a non-linear controller togerher as stated in this document...

Click here for the full document

Interesting paper. But I though you wanted to do GPS navigation and not inertial navigation in GPS denied areas? Isn't that the solution to a different problem than the one you're pondering?

Yes, but as stated in section 4.12 page 161.... I am looking for details of PID approach, I must be doing something wrong... in some cases

4.12 Outdoor Environment Ready
The quadrotor navigation controller utilizes x, y, and z input positions to control
its position and heading. Inside, a sonar and camera yield effective results for height,
position, and heading. Simply replacing these sensors with a barometer, GPS and magneto
meter would yield a system equally capable of controlling its altitude, position, and heading
while in an outdoor environment. Such an approach has been demonstrated before on the
Starmac [31]. Drop in replacements of these sensors would be easy to find and implement,
making this platform usable in various environments.

You probably know this, but I thought that I would mention the difference between the MK and the model described above in 4.12 is that the altitude hold is handeled in the flight board. If you have just a flight board (no Navi, Compass and GPS), altitude hold works. There will not be any GPS altitude information but the altitude hold works with just the baro and ACC Z.

So this is different in that height hold isn't part of the Navi microprocessor algorithms (as far as I think I know).

Hardware_man

Yes, I know.. Thanks for information.. That was a copy paste mistake.. The PID formulas are for navigation but the parameters were for the ALTITUDE control not for Navigation.. I mistakenly copied and pasted that values.....

Thanks for information...

Hi metalman,

I think the main problem for naviagtion algorithms using gps position is the slow update rate of the gps position. Is it something about 4Hz?
I think with such a slow datarate of the copter position u will never be able to reach the exact target position. Because of the strong dynamic behavier of the copter in wind the copter moves too far in between the time of two measurements.
Imagine the copter is close to the target position. Now your controller calculates a vector wich points at the target. But this vector is now used for the next 250ms of flight. This can be enough to let the copter miss the target position. The next calculation of the flight vector could already be right to the opposit direktion. The result is an oszillation or circles aroud the target position.
Did you try very low values for your controller? This might help. I'm not sure if a deadzone around the target position is helpful as well.

What do you think?

In Metalman's opening statement of this thread, he said, "I can't stay over the waypoint as stable as original Mk does."

So it is possible to do, the question is how? My limited knowledge of PID is that your integration time (the I of PID) needs to be longer than your process time. In this case, the 4 Hz update rate??? Then the loop won't "oscillate" or "chase it's tail" back and forth across the way point. But it has to respond quickly if the wind blows it far from the way point. I think that is the D in PID.

Note, Holger does mention some where that he limits maximum number of satellites because after too many satellites, the up date rate gets slower. I think the number of satellites can be more with the S module than with the H module because the uP in the GPS module can read ROM faster than Flash. And I think some have said the 5H is slower than the 4H, maybe part of the reason Holger switched to the S

Can you put a "hook" in your code so that some unused pin on the NAVI uP pulses on each new up date? Then check with a scope and see if you really do have a 4 Hz up date rate?

Hardware_man

Hi All,

It's been a long time that I did'nt inform you back about my progress on building own navigation routines.
After working "HARD" finally we achieved some improvements...


Now these features are implemented and the firmware is shared via the link at the bottom of this page:

1) PID controller is workin 99%
2) we can do position hold within a meter square (very accurate)
3) wind calculations are done, so wind does not effect our stability anymore
4) waypoint navigation works (in sample firmware it is limited to 30 waypoints)
5) no distance limit for waypoint navigation (in sample firmware it is limited to GPS_parameter_distance_limit X 2 = 500 meters)

The code is based on V0.26f of H&I MK.. so their copyright rules run on my published firmware as well...

Use this guideline to fly the new firmware
1) upload the firmware using MK-tool
2) on navi tab of settings do these settings
- GPS_P and D (if possible assign these values to poti's)
- GPS I is not use now, but for a better accuracy we're planning to use the I as well
- GPS P,I, D limits are not used within our firmware so they have no effect of flying characteristics
- on Navi-2 tab , I use none of the settings but the windcorrection.. This is not related with windcorrection anymore, it just changes your maximum stick setting.. Please use a value between 100 and 127
3) Dynamic Position Hold is not implemented, so please be patient that when you trigger PositionHold, it always tries to get that point, even if you move the copter far away from that point manually.



I did some basic tests, but I will be happy to hear your comments of this firmware... Please don't forget to assign P and D to POTI's and play with them while flying...
P : changes the request strength to go to waypoint/hold position
D: changes the brake for applied when we go faster than the calculated speed....

DOWNLOAD LINK IS AS FOLLOWS RAPIDSHARE LINK










PLEASE BE AWARE THAT I HAVE NO WARRANTY, PLEASE USE UNDER YOUR OWN RISK

Did anyone have a chance to test the firmware ? I will be happy to hear about the results of your tests....


-= Metalman =-

Hey,

just returned from my holidays - this sounds like some nice results metalman I would love to test your firmware I'm just very buisy right now - I will definately try it out and let you know.

Greetings

There have been 68 downloads from RapidShare but I think none of you had a chance to test the firmware... I hoper after the weekend I may have some comments on it.... Thanks to all....

topic moved to http://forum.mikrokopter.de/topic-35279.html