Video of AttoPilot Beta Test

[icebear]

Last weekend I managed some beta testing of Dean's AttoPilot and I quickly put together a short video of one of the sorties. Both from the ground and from the test airframe using a small CVS camcorder.

The AttoPilot works well with a Miss2 trainer and I have made 4-5 sorties of 12 WP's so far with good results.

The video is located on diydrones.com here;

http://www.diydrones.com/video/video/show?id=705844%3AVideo%3A6388


Enjoy!

/Icebear

[_helitron_]

Great video Bjorn and wonderful music, like it very much ! Really well done !

//Erwin

[icebear]

Thankyou Erwin!
I hope you have received your unit by now - I will follow your progress... :)
Nice GE track by the way!
/Bjorn

[_helitron_]

You're welcome Bjorn !

Unfortunately not yet a unit but I'm patient :D !

By the way, have you also thought about adjusting your overlay more accurate with the i-Blue ? I was amazed that you've also such a poor resolution on GE at your location but I saw also some miles away from Helsingborg there is a much better resolution :confused: .

Cheers,

//Erwin

[zik]

Nice video icebear!

[dalbert02]

Very nice! I can't wait to get mine!
-dave

[UAVeez]

Very nice! Can't wait for mine too!

[dmgoedde]

Hi guys, I re-opened the thread hee on RCGroups, not sure how it got closed, maybe I did it accidentally!

Still will get units out this week.

[Prashanth]

Nice video Bjorn..

How much of a change in altitude did you notice? was there any during the three laps?

Prashanth

[icebear]

Thanks everyone! Looking forward to seeing testresults from all over the place soon! :)

Erwin - Yes, it IS odd - just outside town the resolution is great but in my area I needed the overlay. I use a track I made by car to correct the placing but I guess 3 GPS coordinates should be better... We have very high res photos on the telephone directory service web page and they even list GPS coordinates. That's actually how I found the exact location of the lighthouse!!!

Prashanth - Thanks! I did not manage altitude logging in the three first sorties (12 laps!) but did another one later and then the altitude settled in nicely around 200 feet after a while (going down from 400 ft when enabled). I will do more tests on this now to see how it performs.

/Icebear

[dmgoedde]

All - I just posted to the testers new firmware... this turns on a logging feature of AttoPilot so a 2nd GPS logger is not needed... during autonomous flight the user can configure a desired data rate from 1 second per data up to whatever you want (5 seconds/data... 10 seconds...97 seconds. really though 1 seconds is fine because of HUGE volume of the SD card).

The log.txt file contains basic GPS info (lon, lat, alt, speed, heading) and also what the AttoPilot is thinking and seeing in autonomous mode (actual heading versus desired heading, actual versus desired altitude, current waypoint #, and positions of all servos as AttoPilot drives them in autonomous mode).

4 units shipped, another 10 mid this week. things are rocking and rolling.

Dean

[Connexxion]

Hey Dean,

Please tell me the IMU-Attopilot is about to be born ;)

Cnnxxn

[Connexxion]

Icebear,

Do I see the wobbling Dean already noticed on his Miss2 in your video?

Looks like it is.

Cnnxxn

[CrashingDutchman]

Nice video Icebear. Can't wait to get mine [acting like I am patient]...

[dmgoedde]

Icebear,

Do I see the wobbling Dean already noticed on his Miss2 in your video?

Looks like it is.

Cnnxxn

Yes - the wobbling is present because Bjorn used the identical firmware I used on the videos I posted.

Firmware I sent Bjorn yesterday adds the gain scheduling which is user-configurable, so wobbling will be a thing of the past.

Dean

[dmgoedde]

Please tell me it's a IMU-Attopilot you're working on ;)

Cnnxxn

In lieu of a full-on Kalman IMU, post-beta AttoPilot (for-sale version) has MEMs sensors included on PCB for wing and pitch leveling that works for moderately stable planes. For example, you can retire the FMA Co-Pilot on your Cularis or Telemaster. (You can retire your secondary GPS data logger as well, plus whatever GPS and sensor suite needed for feeding telemetry transmitter (OSD as well) as well.

Some people have told me that AttoPilot will save them 16 ounces of weight.

[UAVeez]

All - I just posted to the testers new firmware... this turns on a logging feature of AttoPilot so a 2nd GPS logger is not needed... during autonomous flight the user can configure a desired data rate from 1 second per data up to whatever you want (5 seconds/data... 10 seconds...97 seconds. really though 1 seconds is fine because of HUGE volume of the SD card).

The log.txt file contains basic GPS info (lon, lat, alt, speed, heading) and also what the AttoPilot is thinking and seeing in autonomous mode (actual heading versus desired heading, actual versus desired altitude, current waypoint #, and positions of all servos as AttoPilot drives them in autonomous mode).

4 units shipped, another 10 mid this week. things are rocking and rolling.

Dean

What size is the SD card Dean? Is it going to be shipped with the Attopilot or should we purchase one?

Joe

[dmgoedde]

The SD card is included, along with adapter so the micro-SD card can be inserted into a standard SD reader. It is 512 MB, the smallest I could find!! This size is WILDLY larger than you will need EVER for this application.

I estimated that if you logged Lat/Lon/Alt to a file at 5Hz, it would take 47 days to fill the card. Also, a 100,000 waypoint file would be only 2MB... for the adventurous person about 5 million waypoints could be stored on the card (the 100,000 waypoint limit of AttoPilot is a # I pulled from thew air... there is no 'hard' reason for it). These values are what I think I remeber from a few weeks ago when I punched the numbers... but you get the idea.

Also, when I run all the test programs on each Atto before shipping, I use the exact GPS and SD card going out with that unit, AND I test the read/write speed of the SD card. My test routine writes to SD card at about 180kB/second, and reads data at roughly 350kB/s.

[dmgoedde]

All - Icebear had a few nice flights today after I sent him firmware updates that enabled 2 things:

1) User-configration of the autopilot by two .txt files on the sd card. One file controls pin and socket assignments to the physical unit plus settings for distance to be from WP to check it off, radius of holding pattern at the end of flight, and data logging rate. The other .txt control the PID tuning, gain scheduling, etc...
2) automatic logging of all GPS and several autopilot parameters to SD card during the flight. Icebear wrote "Datalogging is also wonderful to have and just too bad I invested in the i-747 logger (anyone want one :)?)."

And also "I ended the weekend around 6PM today at sunset with a 22 WP 20 minute flight (around 8-10 km or 6 miles) with the Atto steering my Miss2 as if on rails..."


http://www.diydrones.com/video/video/show?id=705844%3AVideo%3A7708&context=user

Thanks Icebear! Another 8 units are either arrived to the people or in transit.

Dean

[kd7ost]

Way to go Bjorn. The master tester. ;) Cool video too. You braved the ominous clouds to get the job done. My kinda pilot.

Dan

[_helitron_]

As I wrote on the beta site already, congrats also again Bjorn (and Dean also of course :)) to the great success, the nice video and the great music :) ! Can't wait 'til my beta "Atto" arrives in the next few days :D !

//Erwin

[icebear]

Thanks a lot Dan & Erwin!

It's great to see that navigation and altitude hold now is working and Dean has done a good job of adding features and ironing out some bugs during the last couple of weeks.

I managed another 22 WP route testing today under PERFECT weather conditions - sunny and almost zero wind, but as seen in the video I am racing against time - every day is getting 5 mins shorter and after Sunday it is the end of daylight savings time...

It will be fun to see how things develop for the other beta testers in milder climate and what further tweaking and testing of the AttoPilot can give us.

/Bjorn

[Connexxion]

Another great flight done!

Couldn't see any wobbling anymore. :)
Dean,you're a man who walks the talk. ;)

Looked like the Miss2 did perform cross-tracking or was that just an optical illusion? :confused:

Cnnxxn

[d_wheel]

Icebear wrote "Datalogging is also wonderful to have and just too bad I invested in the i-747 logger (anyone want one ?)."

How much Icebear?

Later,

D.W.

[icebear]

Connexion - yes the oscillation around the path is much less prononuced with the new firware - I will share more details when I have sorted out the data collected. More testing and tweaking is of course needed, but it looks really promising. There was not much wind when I shot the video, but earlier in the day, I could clearly see the crabbing in sidewind.

Dale - I got the unit from Germany a few weeks ago for about 70 euro - that's about 100 dollars. How does 60 dollars + postage sound? Maybe you'll get it cheaper in the US anyhow?

/Bjorn

[icebear]

I just put a new video from my AttoPilot Beta testing on www.diydrones.com

Had to take some video footage from the seaside since we had a great sunny and mild fall day today :)

http://www.diydrones.com/video/video/show?id=705844%3AVideo%3A9588

Enjoy!

/Icebear

[Connexxion]

Hi Icebear,

Do you have any data regarding altitudehold and speed?

Any data regarding recorded and actual flightpath?

I really like the scenery of the autumn colours.

I'm anxious to see/read what the other testers will come out!

Connexxion

[dmgoedde]

Conexxion - Icebear sent me the log files after two flights, one of which is the one mentioned two posts above. I'll let him post the reply with data, but the alt hold was +/- 10 meters from target the whole time.

On the DIY Drones page (or maybe it was a private e-mail to me) he said the purpose of 12 WPs back and forth was to test the 180 degree turns.

[icebear]

Hi Connexion!

Yes, the data was logged by the Atto and here's a picture from the tracking versus planned route.

It's a bit cluttered since there are 5 laps and some climbout/descent in there also but generally you can see that the Miss2 follows path and the plane was stable in turns.

What I really liked to see is that the Miss2 behaves even better with camera and a heavier LiPO pack. It is so light that it easily drifts more in the wind otherwise. The 180 degree turns were kept under control by the Atto (max 10 deg/sec).

Like Dean commented alt hold within approx 10 meters. The drops comes when turning downwind as expected - see enlcosed diagram in Excel.

The speed was between approx 53 kph downwind and 21 kph upwind (windspeed about 4,5 m/s=16 kph or 10 mph)

I am also looking forward to seeing more beta reports! :)

/Bjorn

[Connexxion]

Conexxion - Icebear sent me the log files after two flights, one of which is the one mentioned two posts above. I'll let him post the reply with data, but the alt hold was +/- 10 meters from target the whole time.

A deviation of 10 meters is awesome!

[Connexxion]

Hi Connexion!

Yes, the data was logged by the Atto and here's a picture from the tracking versus planned route.

It's a bit cluttered since there are 5 laps and some climbout/descent in there also but generally you can see that the Miss2 follows path and the plane was stable in turns.

What I really liked to see is that the Miss2 behaves even better with camera and a heavier LiPO pack. It is so light that it easily drifts more in the wind otherwise. The 180 degree turns were kept under control by the Atto (max 10 deg/sec).

Like Dean commented alt hold within approx 10 meters. The drops comes when turning downwind as expected - see enlcosed diagram in Excel.

The speed was between approx 53 kph downwind and 21 kph upwind (windspeed about 4,5 m/s=16 kph or 10 mph)

I am also looking forward to seeing more beta reports! :)

/Bjorn

Icebear,

Could you add an arrow to the (google earth) picture to show us the direction of the prevailing wind at that time?

What's with the wobbling?I thought it was kinda sorted out?

I see a different kind of "slalom" on the left side (over the beach)compared to the right side.
There's more wobbling on the left side,while the right side looks more comfortable for the Miss2.

Connexxion

[icebear]

Connexxion,

Altitude hold works fine, although I think +/- 10m could be improved.
There are variables to fine-tune further and one should remember that this is GPS controlled altitude hold.
The alt hold of the Picopilot for example which uses a baro sensor can keep my Supermiss within 1 meter on a good day.

The wind was from the south and therefore the swaying is more noticeable on upwind legs as you noticed.
From the ground it looks much more stable (and in the video) than on the actual track in GE. It seems my Miss2 has a little more navigating back and forth than Dean in his setup so it can be bit of a trimming/setup issue too.
However I don't see this as a big issue, even if I think it can be tweaked further during the beta testing.

/Icebear

[d_wheel]

Looks great Icebear. How about trying to get one when flying with a crosswind. I would be VERY interested to see what that track would look like.

Keep up the good work guys!

Later;

D.W.

[dmgoedde]

The alt hold of the Picopilot for example which uses a baro sensor can keep my Supermiss within 1 meter on a good day.Icebear

Baro alt control to be turned on in next firmware rev, along with elevator for the turns which will safely allow higher turn rates than 10 deg/second.

It is important to remember that PicoPilot is merely "Alt Hold" not 3D alt control. AttoPilot will have the advantage of using barometer to hit and hold altitude within a meter or so, and change alt at each WP.

[icebear]

Looks great Icebear. How about trying to get one when flying with a crosswind. I would be VERY interested to see what that track would look like.

Keep up the good work guys!

Later;

D.W.

Thanks DW!
I have quite a lot of data so I'll show some cross wind tracking too soon - just need to make a jpeg of it.

Connexion - Regarding alt hold, I forgot to add that it is only a firmware version away as Dean pointed out - the hardware is already there.
I am sure alt hold will be possible to keep much tighter and 3D wp's is a great feature!

/Icebear

[Connexxion]

Dmgoedde,

A little elevator during turns,would be great as I'm fighting altitude loss during turns with my Picopilot at the moment.

The advantage will be the possibility to make tighter turns as you stated and there will be no need to increase throttle to regain altitude after every turn(efficiency).

Cnnxxn

[icebear]

D.W.

As promised - here's an example of tracking in 8-10 mph wind from 180 degrees (East).

By the way - were you interested in my i-747 logger :D?

/Bjorn

[d_wheel]

D.W.

As promised - here's an example of tracking in 8-10 mph wind from 180 degrees (East).


Thanks! Was it flying clockwise around the square?

By the way - were you interested in my i-747 logger :D?

/Bjorn

I was, but have too many things going on and painfully had to decide to pass it up. :(

Later;

D.W.

[icebear]

DW,

Yes, clockwise, and the wind I see now was a little more from NE, but you see that anyway I think.

I know what you mean by too many things going on... I am blessed with "buildning season" starting now so I can catch up! :)'

/Bjorn

[d_wheel]

DW,

Yes, clockwise, and the wind I see now was a little more from NE, but you see that anyway I think.

/Bjorn

Looks like the x-wind part of the program is working. It appears this system will track better than anything I have seen before when the turn rate is cranked up a bit more.

Later;

D.W.

[LukeZ]

Are the squigglies still an artifact from your earlier firmware revision or is that just as good as it's going to get? I understand some of this oscillation can look worse in a graph than in real life, but just curious.


Luke

[icebear]

DW and Luke,

This is from the latest fw update but I must say that the performance is very stable. The Miss2 is quite slow and gets affected by wind easily which can be seen in headwind and x-wind legs.

More trimming can of course be done to the airframe and the autopilot parameters, so I'd say that after just "plugging in" the Atto in my Miss2 this looks really promising.

At the moment the beta testing has just started so I expect there to be some further tweaking in the coming weeks.

Watch this space :)!

/Bjorn

[LukeZ]

Bjorn, thanks for the explanation. All this looks quite promising.

I certainly will be watching this space for more! And I hope to see soon some reports from the other dozen or so testers.


Luke

[BarrelRoll]

In lieu of a full-on Kalman IMU

How many states does the Kalman filter have?
The computing power required is growing as the number of states cubed IIRC.

[d_wheel]

Like Dean commented alt hold within approx 10 meters. The drops comes when turning downwind as expected

/Bjorn

Just curious. I have been seeing references to the "down wind turn" phenomenon for many years. Both in full size aircraft and models. According to all of the most learned aeronautical engineers, an aircraft in flight has no way of knowing if it is flying into the wind or against it. Therefore, when making a turn it will not loose any more altitude when turning downwind than it will when turning upwind.

With this in mind, it would be interesting to see your altitude data correlated with the models direction of flight, and the direction and speed of the wind while making turns.

Mind you, I am not trying to turn this thread into a discussion of the down wind turn argument, just would like to see the data if available.

Later;

D.W.

[icebear]

DW,

That's a really good question and I don't mind you asking!

I need to clarify - I totally agree that the airframe cannot 'know' if it is flying down or upwind - windspeed is the same BUT when turning downwind, the plane has to accelerate relative ground and lift is temporarily affected when turning from a headwind and thus a drop in altitude before airspeed has picked up.
I will dig up some data and post to see if I can 'prove' this soon...

/Icebear

[clolson]

Can you log the servo motion required to hold altitude in a downwind turn versus an upwind turn? Or plot the altitude profile in both turns if you have logging equipment onboard?

Introducing real data will take all the fun out of this discussion, but if we solve this with real data, then we can move on to more interesting theoretical topics.

Personally, I will vote that this "perceived" drop is due to the fact that you the viewer are watching the R/C plane from a fixed ground position. If you were moving exactly with the airmass, then I bet downwind and upwind turns would look exactly the same to you.

[dmgoedde]

Just curious. I have been seeing references to the "down wind turn" phenomenon for many years. Both in full size aircraft and models. According to all of the most learned aeronautical engineers, an aircraft in flight has no way of knowing if it is flying into the wind or against it. Therefore, when making a turn it will not loose any more altitude when turning downwind than it will when turning upwind.


I agree, IF the plane merely had a fixed rudder deflection applied, end of story, no more adjustments. After all, the wind is a moving frame of reference. How would such a plane know anything about the ground?

In the case of using GPS and waypoints at fixed locations on the ground, now you DO have some reference to the ground. Flying directly with or against the wind, there is no vector of off-angle wind to complicate the steering. The plane is simply flying faster or slower ground speed. When the plane has to turn a 90 degree right angle and is now in a cross wind, you have a complicating effect of the PID "I" term having to re-trim in order to crab the plane sideways correctly. I believe this is the cause of wind effect Bjorn is alluding to.

Dean

[clolson]

In the case of using GPS and waypoints at fixed locations on the ground, now you DO have some reference to the ground. Flying directly with or against the wind, there is no vector of off-angle wind to complicate the steering. The plane is simply flying faster or slower ground speed. When the plane has to turn a 90 degree right angle and is now in a cross wind, you have a complicating effect of the PID "I" term having to re-trim in order to crab the plane sideways correctly. I believe this is the cause of wind effect Bjorn is alluding to.


I guess some of that would depend on the specific details of the controller. What I do know, is that if you are flying based on ground track alone, and the wind is significant, the upwind turn happens really quickly and the down wind turn takes a long time to complete.

When you are heading into the wind, small changes in aircraft heading can produce large changes in the ground track heading, and visa versa for the down wind turn.

Take that to an extreme where your plane is hovering directly into the wind (airspeed == wind speed.) A very tiny change in aircraft heading (or wind direction) or even a minor variation in airspeed or wind speed can produce a nearly instantaneous turn of up to 180 degrees in ground track heading.

It's not anything I thought would be a problem until I actually was up flying and trying to navigate based on ground track alone. On the plus side, crabbing happens almost automatically, but on the down side, your ground track heading is much more sensitive to change when flying into the wind versus flying with it.

[icebear]

Curt,

Here is a picture of a typical altitude drop during a turn. I haven't had time to analyze very much, so while there defintely are altitude losses on turns from upwind to downwind I have not been able to rule out that this is just due to the turn itself (not depending on upwind/downwind).

This is interesting - I have lots of data to analyze so I'll report back...

/Bjorn

[John O'Sullivan]

As a long time Free Flight modeller, I have stuck to the principle that the model is flying within its own envelope of air and its altitude is not affected by upwind or downwind turns.

The introduction of external controlling mechanisms such as the ATTOPILOT, with control inputs dictated by global coordinates introduces an entirely new set of turning criteria which are added to the turn and the model is no longer just operating in a discrete moving envelope of air.

I see that the loss of altitude in an upwind to downwind turn could logically result in a loss of altitude. The altitude logging properties of the system will allow development of a much better appreciation and understanding of of the "Dreaded Downwind Turn"

I am looking forward to doing some experimentation on this when my ATTOPILOT arrives. Weather here is cool at the moment (just above freezing), but we fly at least weekly even in the depths of winter.

John

[Connexxion]

Baro alt control to be turned on in next firmware rev, along with elevator for the turns which will safely allow higher turn rates than 10 deg/second.

It is important to remember that PicoPilot is merely "Alt Hold" not 3D alt control. AttoPilot will have the advantage of using barometer to hit and hold altitude within a meter or so, and change alt at each WP.

Dean,

how will the elevator react during turns?
Proportional to the rudder throw or to the amount of altitude loss?

Connexxion

[d_wheel]

According to accepted theory, the things that will affect an airplanes altitude during a turn are air speed and the loss of vertical lift because the wings are not level. If it is only telling the aircraft to turn to heading xxx with a turn rate of yyy then wind direction and speed have no affect. If it was flying level in straight flight and no altitude corrections are input during the turn, it will loose the same amount of altitude regardless of the direction it is turning. This is why I am so anxious to see more data from Icebear. We could be writing an addendum to the physics of flight.

Later;

D.W.

ps If the autopilot is trying to hold the aircrafts GROUND speed constant, we have a whole new ballgame.

[problah]

Couldn't this be resolved by having a forward and reverse facing wind speed sensor (pitot tube)?

You should be able to make the airframe completely aware of up and down wind conditions. You could even possible do side speed sensors as well.

This could then be coded to maintain speed in comparison to knots (as it should), and let the gps decide the nav route accordingly. I would think that would give you consisten speed, and possibly pool in those looser runs back to the square.

[dentompie]

Interesting discussion,
I experimented with stabilization systems (on a delta wing) using both IR sensors and a "normal" IMU. Both systems make the nose of my plane drop when making a turn, but a lot more aggresively when turning downwind.
By compensating the loss of lift (caused by the roll angle) by adjusting the elevator, I was able to overcome this problem. However, it is still not clear to me why a slight loss in lift causes the nose of the plane to drop (and thus causing an even worse loss in altitude)...

Or maybe the PID control algorithm just needs to be better tuned? (Until now I only used the proportional term for controlling the elevator)

[dmgoedde]

Dean,
how will the elevator react during turns?
Proportional to the rudder throw or to the amount of altitude loss?
Connexxion
My plan:
-Pitot for air speed hold via PID of throttle
-Baro (non-pitot) for alt hold via elevator.

Originally I envisioned a feed-forward/feed-back system for the elevator:
1) feed-forward, based on bank angle. Idea is to minimize alt loss before it happens, rather than just react
2) feed-back, based on actual altitude, measured barometrically.

I am a fan of keep it simple in the air. I have a fancy algorithm I wrote for the SCP1000 barometer that simultaneously achieves the higher 9 Hz sampling rate AND precision of the 17 bit pressure normally given by the slower 1.8 Hz sample rate (17 bit, hi res mode). It is a real-time least-squares linear fit of the last N data points. The result is slope, and Y-intercept. Current time (now) is 0 on X-axis, and previous N data are each -N/(1/9th second) on X axis. Slope gives accurate (best fit) dP/dt, and Y-intercept is the best fit of current pressure (at t=0, or now). Result is sub-meter alt resolution at 9Hz update with free info on dP/dt. If you read this far, I am using this level of precision to simply do a very responsive feedback, no feed-forward on elevator control.


As far as other discussions recently on this thread, there are other things that can be done, such as gain scheduling of elev/Rudd/Aileron PID's based on plane heading (ground track) with respect to wind.

On a different topic: I got confirmation that my new assembly Rx/Servo routine works fine with a tester's 2.4 GHz system, so it looks to be 100% universal with any 3V/5V logic Rx, and any pulse order from the Rx into AttoPilot. BTW, the servo outputs from Atto are at 2.5ms seperation (rising edge to rising edge), no matter what the Rx inputs are, so another benefit is easy compatability with 3rd party units that need a servo buffer to alleve overlapping logic signals (FMA CoPilot comes to mind).

Dean

[Connexxion]

Thanks for the time you took to answer my (our) question(s)!

Cnnxxn

[wadiprawita]

My plan:
-Pitot for air speed hold via PID of throttle
-Baro (non-pitot) for alt hold via elevator.

Originally I envisioned a feed-forward/feed-back system for the elevator:
1) feed-forward, based on bank angle. Idea is to minimize alt loss before it happens, rather than just react
2) feed-back, based on actual altitude, measured barometrically.

Dean

Dear dmgoedde,

using this algorithm, how do you prevent the aircraft from looping ? Do you use some kind of elevator travel limiter ?

Regards

-doni-

[dmgoedde]

Dear dmgoedde,
using this algorithm, how do you prevent the aircraft from looping ? Do you use some kind of elevator travel limiter ?
Regards
-doni-

Whoa! I think you have a very old quote from me there. Not that I mind, it is just that Atto is evolving so quickly practically EVERYTHING is out of date now :)

Barometer alone is not enough to stabilize the pitch axis of planes, generally speaking. You really need some sort of IMU for accurate control of the various axes, to ensure truly robust flight. Otherwise, bad things can happen yet the autopilot doesn't really know about it to even do anything about it. Examples would be turning too hard and inducing a tip stall then spiral dive, or a stall or uncontrolled dive in the case of elevator control without an IMU.

With my current system, I am using limiters on pitch, but not a limiter on elevator travel strictly speaking. I never let Atto drive to a pitch target that is outside of a safe range. My data posted about 30 minutes ago had pitch travel limits of +/- 30 degrees. Looping is not a problem, and will never happen in my control scheme, nor uncontrolled dives. Even in the case of engine failure (or Batt cut-off) the worst thing that happens in my Autopilot is that you get a gentle nose-up descent this is nice and slow, kind of like the space shuttle returning to Earth at high angle of attack. Roll is held perfectly flat, and the plane just floats down with nose up a bit.

[octane-link]

Dean:

This is probably old, but your plan for barometric pressure to do alt hold has a problem. A PhD project I saw did that, and when the sun came out, his plane dove to 10 ft above the ground to maintain the pressure reading.

Worth thinking about a GPS calibration or something to prevent this situation from happening!

[dmgoedde]

Dean:

This is probably old, but your plan for barometric pressure to do alt hold has a problem. A PhD project I saw did that, and when the sun came out, his plane dove to 10 ft above the ground to maintain the pressure reading.

Worth thinking about a GPS calibration or something to prevent this situation from happening!Hmmm, I wonder about the exact details of how these other people had the problem...

The barometer I am using is indeed sensitive to light, so it is shielded inside the atto... but I think yoou are talking about the effect of sun heating the Earth, and barometric pressure changing. Yes indeed, the only way to do flights that last more than 30 minutes or an hour is have dynamic updates, of things like Baro against a secondary reference.

[octane-link]

Yes, sorry for the poor details, I was referring to the changing baro as the air heats up.

Perhaps you are running the code differently (and that would not surprise me!) but I just wanted to make sure that was avoided. I brought that up because that is a specific problem we have with that autopilot and one reason my group wants to replace it (possibly with attopilot).

I think we are looking at long flights too, so that would be something we would do.

[dmgoedde]

Yes, sorry for the poor details, I was referring to the changing baro as the air heats up.

Perhaps you are running the code differently (and that would not surprise me!) but I just wanted to make sure that was avoided. I brought that up because that is a specific problem we have with that autopilot and one reason my group wants to replace it (possibly with attopilot).

I think we are looking at long flights too, so that would be something we would do.Barometer cross-referenced against filtered GPS altitude over long time frames is the way to go. Barometer for super high sensitivity, and little drift over short times, and GPS for absolute reference, stable over long periods of time.

[octane-link]

Yeah, exactly how I'd do it! Of course, if Attopilot does that, it will make a lot of people happy on my end!

[dmgoedde]

Yeah, exactly how I'd do it! Of course, if Attopilot does that, it will make a lot of people happy on my end!It is not hard to do, once you define a robust method to determine when GPS data is trustable for altitude.

[octane-link]

I can't wait for this thing to come out, the more I read, the more excited I get!