Autonomous Soaring

[danstrider]

Hey Everyone,

http://soaring.goosetechnologies.com

My goal is to break Joe Wurts' cross-country soaring world record of 140.7mi with no human intervention. This won't qualify for an official FAI record because I can't quite make the 11 lb limit with the current hardware, but surely it will usher in a new era of soaring research and fun autopilot AI!

Thanks for looking and enjoy the website!
Dan Edwards
djedward@ncsu.edu

Update 8-1-07: ALOFT (Autonomous Locater of Thermals) version 1 has flown and it's autonomously soaring successfully!

[shedao]

Well I don't have any money to spare right now due to my own robotic endeavors, but I did give this a mention on my blog: http://brightpanda.com/wordpress/

Anything to help spread the word. It's quite a goal you have and I'd love to see you succeed.

Oh one more thing, perhaps you can supply a parts list on the site of the things you need? Maybe someone will send you some freebies.

[danstrider]

Thanks for the blog posting shedo!

As for a parts list, I have a very well documented need list ... I'll include an excerpt here:

Autopilot Equipment:
Cloud Cap Piccolo autopilot (with student discount) = ~$6,000
Piccolo Dev kit = ~$500
Wire harness = $150
Pressure/Temp sensor node (in case the vario idea doesn't work out) = $400 ??
12v lipo onboard autopilot battery = $90
Piccolario Variometer = $450
TEK Nozzle = $60
Autoplot/Manual RC Safety Switch = $0 (already own)

Aircraft Equipment:
JR 8103 radio (incl servos, rx) = $450
RnR SBXC & winch launch system = $1750
...alternately an Condor Magic 2.9m airframe = $260
(going to have an OS.10FP on a pylon for initial testing, already own)

Record Verification Equipment:
Onboard video camera = $90
GPS Overlay board, BOBIII = $125
GPS for overlay, uBlox SamLS = $0 (already own)
Video transmitter or recorder = $150
Video receiver = $70



An alternate homebuilt autopilot is an option, but I'm much more interested in the controls and programming aspect of thermal hunting rather than spending time developing hardware someone else has already gotten working.

Any helpers? Thanks!
Dan

[John O'Sullivan]

With all due respects, that's a bit of a tall order.
You know, I could badly do with much the same setup myself. If you have any donations left over send them to me. I promise that I'll make good use of them and I'll really enjoy it. Don't worry about the radio, I've already got one.
John

[rclinks2002]

Dan,
What do you plan on using for the "$90" video camera?

[danstrider]

I've been told this is actually a cheap master's project, for what it's worth. Compared to industry doing the same thing (and I can back this number up), it would cost $2.8million or multiple hundreds of thousands of dollars for other research entities. $15k isn't looking so bad anymore....

I'll take the "tall order" as a challenge :-)

Also, after my project is completed, I have a few underclassmen looking into the same type project, so the hardware will likely stay in house for them ... though starting as a modeler myself I really understand where you're coming from. I wish I had someone feeding me old equipment too!

I honestly wasn't sure if I should post this on RC Groups or not. It isn't quite the right forum for strumming up funding, but it is the perfect place to share ideas and to get some quite excellent feedback when things aren't working right. You guys are the king of first-hand experience!

Dan

[danstrider]

Hi RCLinks2002,

I've used a small black widow camera (www.blackwidowav.com) with good results before .... their website is down at the moment, but they have a few options for under $90. I forget if I purchased it at BW or somewhere else, but there is a good quality "bullet cam" that the NC State Aerial Robotics Club uses (see the title image on the T-Hawk gallery: http://www.ncsu.edu/studorgs/ar). I would just essentially duplicate that system.

The video system is low on the list, but would be a very easy way to verify the goal has been achieved ... and it's good visual feedback about how the thermalling algorithm works. I plan to initially have the camera looking back at the control surfaces from the forward fuselage with a mic hookup to the onboard vario for some very rough qualititative evalutation/debugging. The Piccolo gives good telemetry data for quantitative evaluation and I have used Matlab in the past to show 3-d representations of flight paths to illustrate more than just trace charts for elevator deflection, for instance.

Dan

[rclinks2002]

Did you run into the problem of needing a Ham radio license to use those cameras?

[danstrider]

No ham licence needed under 1 watt of transmission power.

[kd7ost]

I can't find the information on it right now. Mr.RC-cam, can you help me out here? To the best of my recollection the highest power output at 2.4 Ghz you can legally use without a Ham license is 10mw EIRP.

Dan

[Mr.RC-CAM]

To the best of my recollection the highest power output at 2.4 Ghz you can legally use without a Ham license is 10mw EIRP.In the USA, an FCC Part 15 approved transmitter {that does not use spread spectrum modulation} is allowed less than 1mW EIRP (radiated) RF power. Even if it is under the power limitations, if a fellow wanted to observe the FCC rules, the transmitting device would need to have a valid registration number printed on it. Usually a ham license circumvents this sort of thing, so it is convenient to have.

[Magician]

Dan,

Glad to see the project is progressing. I might know where to find an SBXC new in the box. We can talk this summer ;)

And I agree with the Picollo, the hardware is done, it works great and the source code is available. Focus on the new stuff.

Good Luck!

Chris

[cobaltdan]

"In the pioneering spirit of the Wright Brothers taking flight in North Carolina, I hope to teach a computer it can use thermals like hawks and buzzards."


We are under contract to DARPA to do exactly this !

We design an build extremely long endurance UAVs. Our LEAPP (long endurance autonomous powered paraglider) uav has a sustainable powered flight endurance of 48-55 hours and we are funded to develop autonomous thermalling capabilities to greatly extend that endurance.

Please contact us to discuss posible synergies.

Sincerely,

Daniel Preston
Atair Aerospace
499 Van Brunt Street
Brooklyn New York 11231
718-923-1709 ph 718-923-1733 fx
www.atairaerospace.com
dpreston@atairaerospace.com

[chrisgood]

If you will be using the stock Cloud Cap datalink for direct pilot control, you will need to use a Futaba FM transmitter. The Piccolo ground control station (GCS) decodes the FM/PPM stream from the trainer connection on a Futaba transmitter in the particular order that Futaba outputs the channels. Please take a look at page 12 of the Piccolo user's guide at:

http://www.cloudcaptech.com/download/Piccolo/Version%201.3.1/Docs/Piccolo%20User's%20Guide.pdf

If you will be using a separate regular RC link for the direct control, the 8103 will obviously work fine. I prefer the 8103 myself (I have two), but the Cloud Cap GCS was designed around the Futaba stream. JR outputs its channels in a different order.

An excellent onboard video camera can be bought for $30 from Mavin:
http://www.mavin.com/index.php?crn=204&rn=343&action=show_detail

This thread shows that others have used it with good success also.

http://www.rcgroups.com/forums/showthread.php?t=479780&highlight=mavin+camera

Chris

[jbeanelliott]

Would not the F1 free flight world record of 192.83 miles be more appropriate for your unofficial goal of distance with no human intervention?

See: http://records.fai.org/models/current.asp?id=308

Category F1: Free flight
Sub-class F1 Open (G) (Glider)
N°102: Distance in a straight line : 310.33 km (=192.83 miles)

Date of flight: 31/03/1962
Record holder: Zdenek TAUS (Czechoslovakia)
Course/place: Plzen - Bory - Holesov (Czechoslovakia)
--------------------------------------------------------------

Sub-class F3B (Glider)
N°156: Distance in a straight line : 226.44 km (=140.7 miles)
Date of flight: 28/05/1988
Record holder: Joseph M. WURTS (USA)
Course/place: Lancaster, CA (USA)

Dan wrote:
>I have started up a new website to showcase my idea and to gather >support for an autonomous soaring UAV. My goal is to break Joe Wurts' >cross-country soaring world record of 140.7mi with no human intervention. I >doubt this would qualify for an official FAI record, but surely it will usher in a >new era of soaring research!

[jbeanelliott]

Has anyone adapted for use with a UAV the $1295 "Themi" thermal centering device sold by Knauff & Grove for fullsize inhabited airplanes?

http://www.eglider.org/index.php?searchStr=themi&act=viewCat

For more details see
http://www.themi.de/Themi%20Centering%20Device.htm

Excerpt from above:

The Themi centering device offers a new tool for glider pilots. Themi measures and computes constantly the values of vertical velocity, position and heading, direction and velocity of the wind. Using these data Themi computes the position of the virtual center of the thermal. During flight in the thermal this virtual center becomes more and more identical with the real center of the thermal.

The position of this virtual center can be indicated with signal lights. Themi uses red LEDs (light emitting diodes) for signal lights. The two signal lights are installed in the normal view of the pilot. He can recognize a signal of the lights without directly looking at them.

In response to a light signal the pilot can steer the glider in ways that will improve his flight around the center.

When a signal light comes on, it can be thought of the illuminated center itself representing the center relatively to the glider. For example the right signal light means that the center is on the right side and in front of the glider.

The signal lights only come on if the present flight situation concerning the center can be improved. If the pilot initiates the right corrections on his own , Themi will not give light signals. Themi is updating the virtual center every second. Therefore Themi always has correct informations of the flight. Themi will always give signals if the flight could be improved.

The following pages show some general facts about thermalling without Themi. Later it will be shown how much easier it is to center a flight with the help of Themi. <snip>

[space_case]

This is a great idea. I've done some work (theoretical) on the subject myself and was going to submit a paper on it recently, until I decided to put some more work into it.

Do you know that NASA did this last year?

I've come across your project before when trying to research the subject. There are not many people publicly working on the problem.

The single largest problem I came across is locating an adequate atmospheric model, or otherwise developing one myself (for the purpose of simulation).

How large is the vertical velocity component? How big is the thermal? How much turbulence? How to differentiate between turbulence and a thermal? Does the thermal move? With prevailing winds? How fast does it move?

Just some questions raised while I was pondering the subject. Some I have answered for myself, others I'm unsure of. Good luck.

edit: I see you have a link to the NASA project on your page, along with a sample thermal.

Can you speak a little about the rationale that went into the thermal model? Any empirical data?

[danstrider]

I must apologize for my delay in replying. My senior design project has kept me incredibly busy recently and I haven't had time for much else.

Lots of questions to answer...

Glad to see the project is progressing. I might know where to find an SBXC new in the box. We can talk this summer ;)

And I agree with the Picollo, the hardware is done, it works great and the source code is available. Focus on the new stuff.
Hi Chris,

We'll definitely talk this summer :-) I'm looking forward to sharing some stories with you. Thanks for piping up!



We are under contract to DARPA to do exactly this !

We design an build extremely long endurance UAVs. Our LEAPP (long endurance autonomous powered paraglider) uav has a sustainable powered flight endurance of 48-55 hours and we are funded to develop autonomous thermalling capabilities to greatly extend that endurance.

Please contact us to discuss posible synergies.
Hi Dan,

I had a chance to look at your website earlier this morning and I will be contacting you soon. My semester is over May 15th, so I will be doing much more auto-soaring planning with undergrad out of the way. I would be very interested to see how we can help each other! I have some thoughts already that I will get together hopefully this weekend. Many thanks for speaking up!



If you will be using the stock Cloud Cap datalink for direct pilot control, you will need to use a Futaba FM transmitter. The Piccolo ground control station (GCS) decodes the FM/PPM stream from the trainer connection on a Futaba transmitter in the particular order that Futaba outputs the channels. Please take a look at page 12 of the Piccolo user's guide at:

If you will be using a separate regular RC link for the direct control, the 8103 will obviously work fine. I prefer the 8103 myself (I have two), but the Cloud Cap GCS was designed around the Futaba stream. JR outputs its channels in a different order.

An excellent onboard video camera can be bought for $30 from Mavin:
http://www.mavin.com/index.php?crn=...ion=show_detail
Hi Chris,

We have actually talked about the Piccolo before on RC Groups. You highly recommended it (as have others, magician above also) and since then I have been reading everything I can find on their website.

Regarding the Futaba/JR issue, I am a JR man myself, but I can't come up with a way around the Futaba input (inverting the shift wouldn't help if they're in a different order). In the interest of Keep-It-Simple-Stupid, Futaba makes better sense. The JR8103 was a hold-over from an earlier system design and I hadn't thought about the issue you brought up in a while. Thanks!

Good link on the cheap onboard video camera. My advisor was very suggestive that onboard video would be good promotional material and would double as verification for the GPS telemetry. I have done a bunch of AP with the NC State Aerial Robotics Club (http://www.ncsu.edu/studorgs/ar) and they have some video equipment if I get into a real pinch.



Would not the F1 free flight world record of 192.83 miles be more appropriate for your unofficial goal of distance with no human intervention?

See: http://records.fai.org/models/current.asp?id=308

Category F1: Free flight
Sub-class F1 Open (G) (Glider)
N°102: Distance in a straight line : 310.33 km (=192.83 miles)

Date of flight: 31/03/1962
Record holder: Zdenek TAUS (Czechoslovakia)
Course/place: Plzen - Bory - Holesov (Czechoslovakia)
--------------------------------------------------------------

Sub-class F3B (Glider)
N°156: Distance in a straight line : 226.44 km (=140.7 miles)
Date of flight: 28/05/1988
Record holder: Joseph M. WURTS (USA)
Course/place: Lancaster, CA (USA)

Hi John,

What do others think about the freeflight vs radio controlled? The reason I chose RC was because it was a specified point-to-point rather than just locating the aircraft once it flew around. If I can eventually manage beating 140.7miles, I won't stop there :-)



Has anyone adapted for use with a UAV the $1295 "Themi" thermal centering device sold by Knauff & Grove for fullsize inhabited airplanes?
I found the Thermi a long time back, but I haven't seen anything about it since then. I'd be interested to know this too!



This is a great idea. I've done some work (theoretical) on the subject myself and was going to submit a paper on it recently, until I decided to put some more work into it.

Do you know that NASA did this last year?

I've come across your project before when trying to research the subject. There are not many people publicly working on the problem.

The single largest problem I came across is locating an adequate atmospheric model, or otherwise developing one myself (for the purpose of simulation).

How large is the vertical velocity component? How big is the thermal? How much turbulence? How to differentiate between turbulence and a thermal? Does the thermal move? With prevailing winds? How fast does it move?

Just some questions raised while I was pondering the subject. Some I have answered for myself, others I'm unsure of. Good luck.

edit: I see you have a link to the NASA project on your page, along with a sample thermal.

Can you speak a little about the rationale that went into the thermal model? Any empirical data?
Hi space_case,

Very interesting to know you have been thinking about it also! It seems to be a fun thought-experiment (to borrow Einstein's term). The process of autonomous soaring lends itself well to further research because the free-flight guys are able to catch thermals with just tailoring aircraft geometry! Any work to automate the process from there is icing on the cake! The best lesson my soaring mentor taught me was to let go of the sticks and watch...

There are actually a few universities starting up autonomous soaring projects. I believe many may be using Michael Allen's (Dryden Flight Research Center) success as a realization there is work to be done in auto-soaring. MSU in particular already has some funding for a full-scale project.

Regarding atmospheric modelling, Michael Allen (DFRC) published a paper on detailing a thermal model: "Updraft Model for Development of Autonomous Soaring." It details the rationale and gives a Matlab code for his suggested updraft model. Contact me directly and I can get you his paper.

You have hit the nail on the head about the variations of real-life thermals from theoretical thermals. The Piccolo has a hardware in the loop (HWIL) simulator which I am planning to use for intial testing of theoretical thermals and then attempting to update the theoretical model with more emperical models from flight testing. Only time will tell how robust the first controller is to atmospheric variations!

Maybe you can share more of your thoughts. This forum is a great place to get feedback and I personally would be very interested to hear your ideas too!


Whew, this is going to be quite a read for everyone! Enjoy :-)
Dan

[jbeanelliott]

Dan,

Are you trying to stay under the 5 Kg weight limit imposed at most of the RC XC soaring contests (like Montague)?

If you are trying to stay under 5 Kg, won't you be over this limit if you use the SBXC with the extra electronics compared to what the XC guys use?

If you are going to be over 5 Kg why not use a glider with a wingspan larger than the 4.1 meters of the SBXC? There have been several 5 Meter gliders introduced recently (reported elsewhere in RC Groups) that are in the same price range as the SBXC. The improved speed and L/D you would get with more wingspan would make your distance goal much easier (of course).

Also, where do you plan on making your straight line distance attempt?
Can this be done in North Carolina? It would seem like you would want to go to an area more optimum for sailplane XC distance flying.

-John Elliott

[danstrider]

After a hiatus to get my undergraduate degree, I am back thinking about autonomous soaring during my off-time!

John, I don't think I'll be overly heavy. I might not hit the 5kg mark since I'll need a power system for initial flights and I'll be carrying an autopilot and additonal batteries. However, I think I can learn from others' similar endeavors and keep fairly low weight.

I'm not 100% sure on a route yet, but cruising down I-95 doesn't seem out of the question. There is a local glider club in Wilson, NC that has some XC guys with a ~10mi course set up already; perhaps I could run that course a few times in succession to practice for a straight-line run.

FWIW, the SBXC may not be optimum, but it should make a good "truck." Someone I knew used to say that the vehicle was just a box with wings; in this case the vehicle requirements are a bit more demanding, but the vehicle must still be a good payload carrier.

I also must say thanks to the recent donor of my flight vehicle. I need to check still to see if he wants some recognition, but here's a preliminary "thank-you!"

Dan

[Ralph Walton]

Any interest?
http://www.nasa.gov/centers/dryden/news/NewsReleases/2005/05-63.html

[danstrider]

I thought I would update folks on my project.

Things are going quite well. A new-in-box SBXC arrived from a friend for a primary test ship. Dean Gradwell donated a Skymelody setup, including a Skypanel annunciator and TEK nozzle. While I am still looking for the key component -- an autopilot -- the project finally feels like it's finally taking off.

I have successfully datalogged the vario signal onto a microcontroller in recent bench testing, so I am ready to flight test this system and look at how easy/hard it will be to make an updraft map of the sky. I also have a GPS I will integrating in over the next few weeks to have position to correlate with the vario signal. I'm still thinking up a good way to show three dimensions plus the vario magnitude (3-d position and a vertical z vector showing magnitude? how on EARTH to do this in matlab???), but I'll post my results once I figure that out.

For now, I have a vario logged flight to show you:
http://www.goosetechnologies.com/soaring//album.php?albumid=4
The vario data should be really neat when coupled with this altitude data. It should show how efficiently different flight paths use updraft lift for altitude.

Dan
http://soaring.goosetechnologies.com

[space_case]

(3-d position and a vertical z vector showing magnitude? how on EARTH to do this in matlab???)

Not sure exactly what you're after, but take a look at these:

mesh
surf
quiver3

[danstrider]

Thought I'd post an update here because I'm restless tonight...

My autonomous soaring project is progressing quite steadily. I have a maiden flight upcoming on Saturday morning of the Autonomous Locator OF Thermals (ALOFT) system. This will just be a vehicle certification and trim flight, but is hopefully the beginning of flight testing for the project.

I'm still working hard to get an autopilot in-hand and the ability to upload my own algorithms to it, but I have most of the harder work coding already done. So yeah I'm focusing on getting the paperwork through. Once it's bolted into the plane though, the tuning should be fairly quick and I can start taking data to tune and tweak the algorithms! I'm stoked with the prospect of finally flying something!

I have a few new photos that should look like light-year leaps from the original few posts in this thread. Enjoy!

Dan

PS: I'll post more updates after the flight this weekend.

[Magician]

Dan,

Looks great! Good luck with the first flight. It's been a long time coming for that big bird.
I hope you're ready for at least a three year development effort :) I know you like a challenge.

Chris

[danstrider]

Well, I hauled the SBXC an hour down the road both Saturday and Sunday only to have mother nature make the decision for me: too windy.

Saturday was a chilly and breezy day. Despite the weather, thermal activity was very good. Several planes brought by members of the Down East Soaring Society went up, up, up! I even managed to spec out the Bubble Dancer on my own without the vario. Brady Baggs brought a winch for launching the SBXC. Unfortunately the two solenoids BOTH fused closed launching another large glider, so we switched to using the backup winch. In the meantime, my pilot and I did a thorough range check. Then, we did two toss-checks to ensure the CG and elevator trims were appropriate. The first went beautifully: the plane was trimmed and glided straight away for perhaps 150ft. The second throw was also good, except the landing was a bit more nose-down; the tail slapped the ground and the rudder hinges all sheared off. Perhaps the plastic was simply brittle from the cold and could not take a less-than-perfect landing. As a result, we packed up the SBXC for the day.

Fast forward 24 hours. Adam and I fixed the rudder hinges to be bomb-proof and crated the plane back to Wilson, NC. The weather was even winder and colder than the previous day. Brady replaced both solenoids, so his winch was ready to go. Dick Proceus, another local member with over 40 years of glider experience, flew a Psychos to test the winds and immediately said "this is not a good day for a maiden." Heeding experienced advice, we packed up yet again.

The next try is in two weeks, Febuary 18th.

This weekend was actually quite a success though! The aircraft glides just fine and is trimmed out ready for a winch launch. I expect it to fly like any other SBXC: clean and smooth. Since the airplane flew on its own weight during the toss-check, I'll consider that a "successful maiden!" It was also extremely useful to do an outdoors range-check and go through the assembly/disassembly checklists a few times in earnest. Lastly, we identified the weak link of the rudder hinging and the old winch solenoids, both of which could have resulted in problems if not corrected before the maiden. All-in-all, this was a very succcessful weekend!

Believe it or not, I forgot to take pictures of the assembled airplane outside, but I did capture the hand-toss on video: http://www4.ncsu.edu/~djedward/SBXC%20hand-toss.avi

Check back after March 18th for hopefully more good news!

[djklein21]

Awesome Dan,
Spending the last summer with Mike should have really tuned you up for this project. Good Luck!

[jbeanelliott]

Very impressive progress on your project.

If you have some time I wonder if you could answer the following:

1. What is the all-up weight presently? (without the autopilot)

2. If you have GPS track data, what is the minimum stall speed?

3. Also if you have GPS track data, what is the minimum turn radius at minimum sink rate and maybe also at other flying speeds?

4. Have you considered launching the SBXC from the ground (ROG) instead of throwing it when winch launching? (I ask because I have a 105 ounce, 4 meter Thermik XL by Valenta Models and my friend says he does not want to throw it for me anymore because it hurts his arm).

Thanks. -John Elliott

[Crashaholic]

Did you make or buy the manual override switch board?

[danstrider]

John
1. AUW: 10.5lb currently .... add autopilot, gps antenna, 900mhz antenna, & AP battery
2. I don't have flight data yet, but in addition to GPS track speed I'll have actual airspeed. At 17.2lb the SBXC stalls around 24kts; at 15lb stall is around 18kts. This is based on others' experience.
3. I don't have flight data yet for turn radius stuff either. Assuming an L/D, bank angle, and minimum airspeed, I can back out the minimum turn radius, but I'm going to have to sit down to do that. Busy Friday today.
4. Yes ROG! I had actually never seen an ROG until this past Saturday. It seems like it has some neat advantages.


Crash
The whole development for the safety switch board is here:
http://www.rcgroups.com/forums/showthread.php?t=360746
I am using one of Jay's boards.


FWIW, I made a GPS datalogger and it's ready to fly, I just need to put it in the air and try! Maybe I'll toss it on another airplane and make sure the thingy works before I head out to Wilson again... gosh I didn't think grad school was going to be this busy!

Dan

[RCAV8R13]

Here are some links for you to check out.
http://www.dfrc.nasa.gov/Gallery/Movie/Autonomous_Soaring/HTML/EM-0090-01.html
http://news.nationalgeographic.com/news/2006/04/0407_060407_airplane.html
http://www.technovelgy.com/ct/Science-Fiction-News.asp?NewsNum=471
http://www.cloudcaptech.com/gallery.htm
http://www.airtalk.org/nasa-model-uses-thermals-vt46236.html
Enjoy.
RCA

[danstrider]

I took ALOFT back out to Wilson to make sure she flies correctly before I try adding autopilot hardware. A couple guys from Down East Soaring Society came out again, Dick Proceus and Brady Baggs, to help with the maiden flight; both gentlemen have flown lots of XC in the past with large gliders, so I really appreciate their input. My pilot and I made the drive from Raleigh to Wilson a few hours before everyone else was to show up. Adam Propst and I set up ALOFT and then flew a few different personal airplanes to get our fingers warmed up and assess the day's conditions. It was windy and turbulent low, but conditions were slowly improving.

After a complete 4-quadrant range check and voltage double-check, ALOFT was ready to go. The foursome debated how best to launch, ROG versus hand-launch, what launch preset to use, how much wind was too much wind, and so on. It was great discussion and I learned quite a bit about large-scale flying and setup. Onward to the good stuff...

We hooked up ALOFT to the winch line, Adam stood with the transmitter, Brady stood with the plane in hand, Dick managed the pedal, and I stood to the side with a big knot in my stomach. (As a side-note, I decided not to be the pilot not because I think I will have my nose in a laptop for autonomous testing, so I have handed the responsibility to a great pilot buddy from school, Adam). Then we all gave the conditions our thumbs up, Dick started tapping the winch to build up line pressure, and Brady let go. ALOFT pulled forward and rotated nose-up. It flew up a good hundred feet or so and then the line broke! We noticed the plane and parachute went backwards, not where they were supposed to go! Adam calmly pushed back over and continued the test flight.

This plane absolutely floats. With an L/D near 30, even with the meager 100ft altitude from popping off the winch, Adam flew circles around getting the trim in check and getting used to the large aircraft turning performance. Finally after a long 30 seconds, he set up for a landing approach and put the plane on the ground softly. Yay, ALOFT flies!

We flew 5 more flights, adjusting the CG a bit more forward and changing the elevator trim. The winch line breaking problem was fixed, attributed simply to too much elevator up-trim on the first launch. We also had to get used to flying the airplane faster than we expected, but it sure moves nicely once up on step. Landing also were tricky at first, but are much nicer now with a new technique of using full flaps and holding the nose down on the approach and then easing off the flaps to stretch the glide as needed. It looks beautiful in person seeing the airplane coming down like that.

So there is the story and the verdict: ALOFT flies just dandy. Next step, moving onward to the autopilot installation and tuning, once it arrives (nearly completed with the loan paperwork to get one in my hands!). Then, it's time to finish programming up the algorithms and start finding some thermals!

[danstrider]

Thanks for the links RCA!

[danstrider]

More news ... I received my loaner autopilot! This project is really coming together now :-)

Dan

[Gary Mortimer]

very cool, well done

[danstrider]

Big update: *it works* :-)
http://goosetech.homelinux.com/soaring/journal.php?sxEntryID=37

Dan

[LukeZ]

Dan I just read your entry and I have to say, that is incredible news. I can see this having considerable, wide ranging applications. I'm definitely looking forward to hearing more.

Great job though, I know you've been working on this for years, since before the UAV forum was even around seems like. Really nice to see people succeed. Congratulations!


Luke

[jfrickmann]

Yes, congratulations Dan! I was there for the first maiden when the rudder broke off. It is great to see the project making good progress!

Jesper

[slopeHappy]

Dan,
I would also like to congratulate you. It is exciting to see the results of your hard work start to show and I am also eager to see what will come from this. I am glad to have been part of your journey and I see this as a good example of how NASA's education office can work to boost our nations future engineers and researchers.

-Michael Allen

[danstrider]

Thanks to all you guys who have been supportive and to those who have been there to help find the pieces! Mike, you're right about the loan boosting my engineering skills, not to mention all the folks younger and older who I have showed that this project really is something they can relate to! Kudos to NASA helping out the little guys!!

Don't count me done yet ... the best is still to come :-)

Dan

[Magician]

Dan, I'm starting to have SBXC withdrawals :cool: Can't wait to see what you have to fly next.

Michael, great work on the NASA effort. Hopefully we can keep building on the capabilities. Enjoy the great NW and say hey to my old flying buddies at PASS.

Chris

[danstrider]

Hehe, funny you should post ... I'm working on a ton of fixes based on our last flights as you posted! It's great to have the autosoar files from Aug 1st to play back b/c everything is clean and logical. My list of updates though is already a page long, ack!

I did realize too that the speed optimization could be scary complicated if the autopilot could also command the flaps like was suggested. All the sink polar we did would have to be duplicated for a range of flap settings, not to mention the speed algorithm paging through the sink polars for each flap setting then picking the speed ... I think a single clean configuration is how it's going to be for a while. On the other hand, looking at the commanded speed to fly, looks like that will make a difference as far as getting across the sky quickly. From our long 1.5hr flight Aug 1st, we flew through ballpark 50 miles of air with only just the single speed set; I gotta believe we can make some big distance improvements flying faster.

Mike, I spoke with Jay and tried to convince him soaring was good for him to consider, but he wasn't yet sold on the idea. I guess he's so used to flying in a straight line or trying to stay over an area of regard that he was asking questions about how soaring would take the aircraft away from its intended mission. If you can figure out that part, he might be up to try. Then again, it's a hard argument with a gas engine. The new aircraft at the show was indeed very nifty btw! (even though it had an engine ;-)

...

* Cross-Country stuff *

As a side note, for the SBXC with its glide ratio of 25:1 at 12.5lb, for a 10 mile course only 2112ft of altitude is needed and at max L/D speed of ~23 knots would take 37 minutes. Of course that would be silly to fly exactly with that profile, but it does give some perspective on the feasibility of a long XC flight.

For a 150 mile flight, just multiply all the numbers by 15. The plane would need 31,500 ft of altitude gain from thermals and 9.25hr of flight time. This time does NOT include time spent thermalling. If the average climb rate is 300fpm (BEST rate I've captured), that's an extra 1.75hr of thermalling time for a total day's flight of 11hr.

In other words, flying at max L/D when not in a thermal is a poor choice, hence needing this work on speed to fly. ... let me clarify this position: for pure distance max L/D does make most sense. However, at 11hr of flight time, making a 150mi goal is going to start having daylight issues. Thus, working on flying faster is a necessity.


... back to work!
Dan

[jfrickmann]

Hi Dan,

Why not set the flap position to whatever is optimal at any speed and then "profile it out" from the polar by using the polar conditional on those flap settings?

SinkRate(Speed, Flap) = SinkRate(Speed, Flap(Speed)) = F(Speed)

Jesper

[danstrider]

Set the flap position to whatever is optimal at any speed
Hi Jesper,

I'm not sure how to find out the optimal speed for a given flap setting without doing a new sink polar for that particular flap setting ... which involves taking about 10 points for each flap deflection. Once it's done, you're right it's just a simple lookup table.

Is there an easy way to relate a flap deflection to a change in the baseline clean sink polar?

Assuming perfect knowledge of the sink polar for each flap settings so speed to fly is solvable, then what critera would constitute optimum flap setting? Perhaps a better L/D or better min sink depending on the desired mode ... this actually would be a really neat set of research in its own right.

Ideas? I'm stumped.
Dan

[jfrickmann]

Hi Dan,

Actually, I was thinking optimal flap for a given speed. And that would simply be the flap setting that gives the minimum sink rate at that speed. To find that, you would need to "bracket" the minimum sink rate inside three different flap settings. That is, the middle flap setting gives the lowest sink rate of all three settings. Then, fit a 2nd order polynomial to get the estimated optimum. You would also want a polynomial regression of the speed to "smooth" the polar wrt. speed.

If you wish, I will be happy to help analyzing data with such a model. That's what I do for a living (when I am not hanging out here at RCGroups ;) )

[jbeanelliott]

You might be interested in the article titled "Pitch to Fly as a Soaring Tactic" which discusses the use of pitch-to-fly in place of speed-to-fly for full size sailplanes. Although the article is more related to the use of the elevator setting as opposed to the flap setting I think it would be relevant. I was recently looking at my copy of this article but I seem to have misplaced it.

I think back issues of Technical Soaring may be available online at:

http://soaringweb.org/Soaring_Index/Technical_Soaring/Technical_Soaring_issue.html

Volume XVII, Number 4, October, 1993 (up to table of contents)

Michael L. Steinberger, Pitch to Fly as a Soaring Tactic [Performance\Speed-to-Fly], page 101

-John Elliott

[danstrider]

I was digging through my old threads and thought I'd post an update...

In September 2007, I had a 10.1mi XC run with the plane. The auto-soaring algorithms made 7mi before getting too low and requiring my manual pilot to take over and find a low thermal. We turned the auto-soaring back on after finding that thermal and it completed the remaining 3mi of the course.

In April 2008 (last Friday), I posted a 29.4mi XC run with the plane at an average course speed of 25.9mph. The auto-soaring was in complete control the entire course, with my involvement just to hit the "start" and "stop" buttons. We could have made a longer distance, but batteries were getting low.

Thanks for the continued readership and interest!
Dan
http://soaring.goosetechnologies.com

[djklein21]

Man Dan that is Awesome. The guys at Dryden must be really excited for you.

:)

[JayFrancis]

Nice work Dan!

[Neil Stainton]

Congratulations, well done!

Neil.