Thread Tools
Aug 18, 2007, 07:17 PM
Registered User
dmgoedde's Avatar

What you are looking for


All - I have a real and working autpilot that I intend to sell as a ready-to-go and easy to setup unit (not a rough development system for tinkerers), and it is a serious advancement over what is offered by anyone else out there for hobby-level systems, plus it will be cheaper that what you pay them for a full system, and even my Rev0 development system is half their total wieght - about 1.6 ounces including GPS and antenna. Specs: Up to 100,000 waypoints, independent altitude at each WP, Lat/Lon resolution 0.00001 degrees (about 3.8 feet) and the WPs can be up to 1.5 million miles apart (thus no limitation on Earth). Teach the unit by manual flight for about 1 minute, after which it does data fitting and stores the important flight charactersitics to a .ini file on a micro SD card (the same card that holds the waypoint list as a text file). Making the waypoint list is easy: click out a path on Google Earth then save the .kml file and run it through a simple program to generate waypoint.txt. The fly-teaching process trains servo direction, approximate center positions, and response slopes (gains). This system is only currently for stable motor glider types. There are failsafes to handle loss of GPS lock, and the system performs a return to launch and holding pattern if there is a problem, otherwise the user can define a final holding pattern at a site other than the launch site. size 3"x1.1"x0.5" and unit literally plugs between the R/C Rx and servos. See attached pictures.
Last edited by dmgoedde; Aug 19, 2007 at 07:30 PM.
Sign up now
to remove ads between posts
Aug 19, 2007, 02:04 AM
Registered User
icebear's Avatar
That sounds really interesting dmgoedde!

Is the firmware upgradeable in your unit?

How do you manage wing levelling (Co-Pilot or integrated?).

Need a beta-tester with small electric UAV experience in Europe ?

/Icebear
Aug 19, 2007, 03:26 AM
Weekend flyer
Munin's Avatar
Hello

In relation to what MacBoffin said I say run your product line from an European address; low tax area like Isle of Man. Over here in Europe we are some what more liberal when it comes to UAV technology.

Send a unit to IceBear for pre-market release. IceBear is a reliable person and he is located in the ever so popular country of Sweden.

/Munin
Aug 19, 2007, 06:04 AM
Multicopter/FPV/UAV/HAM
_helitron_'s Avatar
Hi, I'm also very interested in this device !

/Erwin
Aug 19, 2007, 12:47 PM
I second the motion on Icebear... He is a fine, responsible gentleman and an experienced researcher that may be of great help. Besides, Swedish extremism is an oxymoron.

Iíd like to suggest starting a different thread to begin public Beta testing. It will be a big help to you as developer to have a Home Thread.

Speaking of Home Threads, readers of this thread may be interested in my PicoPilot start to finish blogs at:

https://www.rcgroups.com/forums/showthread.php?t=511550

https://www.rcgroups.com/forums/showthread.php?t=631757

These threads are similar to what UAVeez is seeking to compile (I believe).

Jeff (aka Workshop when Iím home at the workshop)
Aug 19, 2007, 07:23 PM
Registered User
dmgoedde's Avatar
Yes - I am very interested in beta testers!! I was throwing around ideas how to do it, such as tester can keep the unit provided they pay shipping and cost of parts (about $135 US for the parts), and they agree to wring the model out and provide feedback. Or, perhaps a bit below cost, or I'll pay shipping. Either way, you would be getting an awesomely functional hobby-level autopilot for under $200! If you decide not to keep it I would refund whatever money and say "thank you for your time".

Firmware - yes, upgradable. Besides my ability to re-burn the eeprom via a USB cable, I'm working on a way to have the firmware upgradable via file from SD memory card. I have a scheme, haven't tested that yet. If my scheme doesn't work, I have two options for the beta testers: 1) More desirable is that you plug unit into your PC via same type USB adapter I use, then burning the binary file to eeprom is trivial (I e-mail out new versions), 2) less desirable is that I do what UNAV does and have you ship the units back to me and I do it. I really prefer the SD card method, so I'm focusing on that.

To answer some more questions: The pictured unit (that I could make copies of and send out in a matter of days) is only for stable motor-glider types, like Hobby-lobby Miss2 (plenty of dihedral on wing). It has flown the Miss2, and even the tiny little "Tipsy" - a 9 oz trainer. This autopilot does not do wing leveling. The Autopilot w/ GPS weighs a bit over 45 grams and draws about 50 mA. The shown unit it handles 5V logic from the R/C Rx, but given that I'm using a 3.3V processor, I can make some easy parts swaps and it will read 3V logic just fine (no need for level shifters). OK that's sort of a tangent, however my point is that the unit is flexible. when I get home tonight I'll post a more formal list of specifications to help people decide if it will work for them and if they want to beta test. One of my main goals is to eliminate goofy limitations I percieve on other "hobby" autolpilots. I suppose the only real limitation of my pictured unit that it is intended only for inherently stable planes with polyhedral wings. I'd like to be the Henry Ford of these things: to make it easy for hobbyist to cheaply get into this if they aren't interested in tinkering with making electronics/ development-troubleshooting/programming.

I do have a 5 DOF IMU I developed that employs temperature-stabilized rate gyros. I have successfully used the IMU to achieve and hold pitch and roll flat with respect to gravity on an aileron plane with absolutely no dihedral and with the wings low on the fuselage (not a shoulder wing). If you look carefully on my pics you may see an empty socket on the main board right next to the occupied GPS socket. This unused socket is for my 5 DOF IMU. I have litho masks and plenty of PCB and components to crank out more copies of the same design on demand, so small-scale production is no problem on an asap timescale. I simply have not yet integrated the IMU programming code with current autopilot pictured above. The parts for IMU board is about $70 and it is 12x25x4mm. See attachment showing the IMU board.
Last edited by dmgoedde; Jan 04, 2008 at 12:28 PM.
Aug 19, 2007, 08:25 PM
Stupid gravity!
docphi's Avatar
Wow! Sign me up to beta-test. Great job! How about an integrated OSD to see the data for FPV.
Aug 19, 2007, 10:38 PM
zik
zik
Autopilot guy
zik's Avatar
Nice SMT soldering work. It's finicky stuff!

Also interesting to see that our projects are using a number of similar components.
Aug 19, 2007, 11:09 PM
ios
ios
Registered User
Very impressive indeed,Ö Iím also keenly interested in beta testing!

Iíve got a 3m powered glider ideal for the task.

Iím really impressed that the Rx plugs into your autopilot, reducing the need for additional wires b/w the two units. Which Rx are you using? Also, is the altitude intended to be controlled by the powerplant or the elevator? And what if the altitude isnít attained i.e. short distance between waypoints, large elevation difference ? What failsafe operation does the autopilot perform when GPS lock is lost? (handover to rc / dead reckoning navigation back to luanch / straight and level flight/circling ??)

Keep up the great work, looking forward to hearing from you & your progress. These are exciting times arn't they !!
Aug 20, 2007, 12:39 AM
Registered User
icebear's Avatar
Thanks for your trust in us Swedes Jeff and Munin!!!

It looks like a really nice unit you have dmgoedde and the possibility to add a 5 DOF IMU sounds exciting.

I have a new Miss2 testbed sitting here and winter is still several months away, so I'd be happy to do some beta-testing.

/Icebear
Aug 20, 2007, 12:58 AM
Registered User
dmgoedde's Avatar
Wow - it's cool to get so much response! Thank you.

A 3m glider would be perfect to test. Although my unit worked in the tiny hobby-lobby "Tipsy" (9 ounces) it flies better on the bigger 30 ounce "Miss2" (old-timer, polyhedral wings, rudder/elev/throttle). A larger 3m glider should be rock-solid and with LiPos the duration could really exploit my autopilot's ability for essentially unlimited waypoints.

The R/C Rx is a Hitec Mini 6S 6 Ch. 72Mhz Micro Receiver (Auto Shift Select). I removed it from the case and shrink-wrapped it (mostly to hold the crystal in place). It is fairly cheap, around $35 and boasts a 1 mile range.

Current firmware controls altitude by the powerplant. If altitude isn't met no big deal - as the waypoint "satisfy" criteria depends only on being within a specified distance tolerance of the target wP (r^2 = X^2 + Y^2 pathagorean theorem). right now I'm using 200', though I used 50' for the Tipsy. By the way, I don't like controlling altitude by power plant because a "properly" setup airplane should have thrust line adjusted so that nose doesn't rise or fall with changes in motor power (i.e. throttle should not cause plane to climb). With the Miss2 the thrust line is pretty far below the wing, so increasing power is an effective way to control altitude. I'm currently hammering out the firmware to use my on-board barometer for elevator control. It's also important to note I'm not using a pitot tube for air speed control. I just make assumption that X amount of power correlates to Y amount of climb rate (positive, nuetral, or negative). I ignore airspeed... there is no provision in my waypoint.txt file to specify speed. So in short, if altitude isn't met, the autopilot still drives the airplane to the waypoint, hits within a "satisfy" tolerance, checks off that WP, then starts targeting the next WP, and altitude is a seperate control PID algorithm. I.e. the autopilot does its best to hit the altitude targets. The proportional control band for the P term is 100 meters right now, and the I term trims out throttle rather gently. What I'm expressing is that the plane hits altitude and locks on very well without oscillating over/above target.


Failsafe when GPS lock is lost: Yes, this firmware checks for valid GPS lock at 5Hz rate, and if lock is lost the current firmware would just continues flight in a straight line. This is one thing I haven't optimized yet, but here is the solution I'm working on (and I think it will be pretty much optimal): If GPS lock is lost the plane maintains current altitude via barometer, and enters a slow turn of approximately 10 deg/second. due to no GPS lock, the actual turn rate won't be known, but the plane will execute a turn based on known response factor of rudder from the manual training constants that were originally taught (and possibly refined at regular intervals during even autonomous flight). When GPS lock is re-gained, the plane goes on its merry way flying through the rest of the waypoints. In a real-world situation it might be nice to have telemetry so you know what's going on and where the plane is so if lock is never re-gained you can go meet the plane and recover it before batteries give out. One idea I'm batting around for the future is telemetry via cell-phone modem. These days with so much cell coverage, even in rural areas it would have practically infinite range. Sparkfun.com sells a few cell phone modems though they aren't cheap, but still cheaper than losing a plane.

the paragraphs above are a mixture of current capabilities with a smattering of items on the near horizon that will add a lot more ease of use. I'll make another post on current capabilites for the interested beta-testers, and a list of capabilites I'll have within the week.
Aug 20, 2007, 01:06 AM
Registered User
dmgoedde's Avatar
For those not familiar with PID control, what I meant by "proportional control band is 100 meters" I don't mean that altitude has a 100 meter uncertainty for my autopilot. The 100 meter control band means that if craft is +50 or more meters from (above) target, then motor power would drop to 0% of Proportional control band, and if -50 meters or less (below) from target the proportional control goes to 100% of power. right on target power is at 50% of control range, and for fine-tuning to target the I term of PID self-trims over time. Eventually the plane is within GPS or barometer resolution of target altitrude.
Aug 20, 2007, 01:36 AM
Die noctique
dmgoedde,
Your autopilot sounds great and I have a couple of questions.

Could this be used on a Nitro Plane?

And will there be a way of it activating third party camera triggers? I ask this as I have a "Blip" trigger for a Canon 640 and was wondering it there would be a way of it flying to a way point then once reached fire the camera (via the Blip) for x seconds before it goes to it's next way point?

Also I'm VERY new to UAV's and new to RC in general so would the operator get there plane/glider in the air and then the autopilot kicks in from activating a button on the TX or do you just start the model and off it goes? Finding the whole subject very interesting BTW
Aug 20, 2007, 02:01 AM
Registered User
dmgoedde's Avatar

Specs for those interested in Beta testing


Potential Alpha or Beta testers:

Alpha (current) unit that I can send today:

1) Up to 100,000 waypoints, stored on microSD flash card as a .txt file
2) Waypoints have no limit to distance between them (actually 1.5 million miles)
2.1) waypoint resolution is 0.00001 degrees, or about 3.8 feet.
2.2) For the trig calculations for navigation, the autopilot corrects the physical size of longitude as a function of degrees latitude above (or below) the equator. for example, where I am (33N and 111W) longitude is only cos(33deg), or 83.9%, as big as latitude.
2.3) there are no hemisphere restrictions to where this autopilot can work, in terms of the math employed in my firmware. As long as you can see enough satellites to get a valid lock, then it is party time.
2.4) My firmware is only setup to use the ETek EB-85A GPS. I get quick locks on 9+ satellites even in my basement and the data rate is 5Hz.
3) Altitude individually specified for each waypoint
4) checksum is used for each waypoint on the .txt file, if read error occurs, flight aborts and returns to launch site at +100 meters AGL, and error data is written to an "error.txt" log file on the SD card for troubleshooting later.
5) Upon power-up of autopilot the first 40 WPs are loaded to processor RAM. During the flight waypoints are loaded to RAM as needed in blocks of 20 from SD card, otherwise the SD card is not normally "mounted" during the entire flight, just as-needed to retrive waypoints.
6) By default, a holding pattern is created and exists in RAM which is an octagon of points circumscribed 100 meters above launch site. If there are any WP read errors during the flight the plane returns to the launch site holding pattern and circles as an infinite loop until user turns on the Transmitter takes control and lands the plane manually. During creation process of the waypoint file, before checksum program is utilized you can insert an "H" or "h" at beginning of a waypoint line and the autopilot will then do the trigonometry and make up a new "octagon" holding pattern circumscribed around the coordinatesof that point... then you just go out and meet the plane at that holding site to take manual control and land it (see #7 below).
7) Upon power up, the autopilot reads an auxillary reciever channel and memorizes that channel's value - and assigns this to mean "autonomous" flight mode. If the R/C transmitter is on and that auxillary channel is within +/- 50 microsecond pulse width of the memorized position, then the autopilot executes to the waypoint list in autonomous mode. If that channel is outside the +/- 50 range then flight is "manual" and the R/C Rx signals for for the control surfaces and throttle are patched through to the servos/speed control. If the transmitter is "off" then the autopilot is in autonomous mode. the way I fly is to manually fly the plane about 20 meters AGL get it to a good trim (see # 8 below), then switch to autonomouse, and once I'm convinced plane is flying itself fine (it alsways does) then I shut off the transmitter and collapse its antenna and just watch the plane fly. It doesn't matter if the plane goes outside transmitter range. When I want to abort mission or at end of mission i retrieve plane by turning on transmitter, taking control and landing the plane.
8) When you switch from manual to autonomous, the autopilot uses the last manual flight values for the control channels as an approximate center trim position. Of course by way of PID the autopilot has the freedom to re-trim itself as-needed to perfect the flight control and "nail" the waypoints. When I switch from manual to autonomouse, instead of just shuttind off the transmitter I first set the auxillary channel to "auto", then shut off transmitter.
9) I create waypoints in a really easy fashion that anyone with Google Earth can do: create a path of points and save that path on your desktop as a .kml file. you can specify absolute altitude in GoogleEarth for all the waypoints, or later hand-edit the list if you want specific altitudes at different waypoints. The .kml file is ran through a simple utility for cleanup to a .txt file. This .txt file is imported into an Excel macro I wrote. It is here that you can specify individual altidues or add an "H" at beginning of a waypoint row to mark this as a non-default holding pattern location. Excel does some text formatting, then you save the .txt file and run it through another utility that calculates an XOR checksum for each waypoint line. This is now the final flight-ready "waypoint.txt" copied onto the microSD flash card and inserted into the socket of the autopilot. The autopilot reads this file and executes that flight plan. I DO PLAN to make a single and simple utitility that takes a .kml file and in one step generates the "waypoint.txt" file. Another even simpler possibility is outline below for the "beta" version.
10) elevator stays locked at last position from manual flight. It is important to achieve good trim of elevator before switching to autonomous. This works just fine for stable motor-glider type planes when the wind is not too strong.

Firmware Beta1.0 unit that I could send roughly next week:
1) Have the autopilot directly read .kml files from Google Earth.

Firmware Beta2.0 unit in about a month
1) barometer control of elevator
2) flight path data logging to SD card, as a .kml file, so after the flight you can easily load actual flight performance and see it in 3D using GoogleEarth, and compare planned versus actual performance of flight plan.

Hardware Beta3.0 in about a month:
1) unit still has micro SD socket, however also a mini-USB jack and interface chip - result is that you plug USB cable from laptop to the autopilot, and the autopilot comes up on your computer as a drive that you can drag and drop files into instead of having to remove SD card to do file transfers.

I love this stuff! I love how you can learn stuff over the internet. I didn't know anything about SMD or microcontrollers at the beginning of 2006!
Aug 20, 2007, 02:12 AM
Registered User
dmgoedde's Avatar
Badllarma:

good questions. One thing about my autopilot I neglected to describe is that yes I made provisions in the waypoint.txt file for triggers, through my current firmware isn't setup to use them. the idea is to use them to trigger a camera, turn on a cell phone modem and place a call, or actuate a servo to drop something.

Yes, this would definately work on a nitro plane. I might have to add some code such that you manually train the throttle range by driving transmitter stick from low to high... that way the autopilot can't go low enough to shut off your plan's engine The autopilot would then limit its throttle range to within your specification. When you want to land you take autonomouse control via the auxillary channel, land it then shut off the engine by decreasing trim on the transmitter. Actually, any scenario of control you can dream up I can code.

Having a beta tester with a gas plane was my intent on a recent triip home to Indiana though it didn't pan out. My favorite uncle in the world (my inspiration for R/C planes since I was a kid) has an old "Goldberg Eaglet 50" with a 20 nitro. His radio equipment is form the late 1970's and is weird with more than three wires to each servo, so needless to say we didn't do any testing on that plane!

Would you be interested in beta tesing on a gas plane? What airframe do you have in mind? Remember, my alpha and proposed beta versions are only for very stable trainer/polyhedral glider types. This autopilot has tons and tons of features that I think make it really stand out, but it essentially just steers the plane around the sky like driving a car or a boat; it can't handle planes with flat wings.


Quick Reply
Message:

Thread Tools

Similar Threads
Category Thread Thread Starter Forum Replies Last Post
Good place for low cost 3 cell lipo... dacaur Foamies (Kits) 2 Oct 17, 2005 09:47 PM
Head's up - New Low-Cost 370/400 Brushless Motors! Peter Khor Power Systems 447 Feb 14, 2005 08:39 AM
low cost bearingholders for the hornet's tailblades mattijs321 Micro Helis 3 Oct 21, 2004 03:22 PM
Low cost charger for 1-3 Lipo cells bz1mcr Batteries and Chargers 13 Apr 16, 2004 01:07 AM
hotliner on a low cost power system ? oded mazor Electric Sailplanes 12 May 07, 2002 03:42 PM