Hobby UAV projects

[joelhaasnoot]

If anyone is involved in building a UAV, whether in the past, the future or currently, please post an outline here. Some ideas for information are:
- Radio Link / Modules
- Imaging
- Control Station
- Programming
- General model information
- And anything else you wish...

Oh, and pictures would be nice...

[radiohound]

Ok, my uav project is somewhat in the future as you can probably tell from my lack of airframe and missing electrical components. I have some pictures of my progress, and here are my thoughts on the hardware/software.

Here is the plan: To the left is a thermopile board for horizon and attitude sensing based off the Paparazzi UAV Project. They have some great information at their project website. This board gives two analog outputs, and using a little software, the angles can be calculated.

I love the Paparazzi plan, but they use amtel processors, and my preference is for PIC processors.

The board in the middle is the main board. It contains two processors. The larger one is a PIC16F876 28 pin processor with 5 a/d converters and lots of in/outs and 8k of space. It looks like it has the same footprint as the PIC18F2320, so that should work too. I plan to have this processor compute altitude from the visible MPX4115a pressure sensor and a 15 bit a/d converter, and speed from the MPX5050 (empty space on board). It may also serve as a servo controller, otherwise it will send serial data to a separate servo controller. It will read the output of the attitude and horizon sensor, and receive gps information from the other processor. It will also control the data transciever shown at right.

The other processor will pretty much be busy reading data from a GPS and sending this to the main processor. This is a 4k PIC16F648A. It does not require a crystal, and can run 4 mhz internally.

Both processor are in circuit programable using two pins.

The Maxstream 900xtend module looks to be a great fit for what I am trying to do. I have heard many people comment on the Maxstream modules and it would seem that either you love them, or - you hate them. They are capable of 14 mile range within line of sight with a standard antenna. They allow high speed transmit/receive with minimum time in between. It weighs in at 0.6 ounces without the antenna. It has 10 available "channels". It is actually frequency stepping with 10 different modes.

To save weight, cost, and do some pretty cool RC things, I plan to use the tx/rx module for radio control, as well as telemetry. It is fairly simple to send serial data over the modules. These can pretty simply be sent over the modules, then to a servo controller much like Parallax's serial servo controller.

One of the cooler capabilities of such a system would be that you could store model information - ie servo trim, rev/std, etc in the receiver on the model rather than your transmitter. Think about it, how many times have you almost (or really did) leave your model on your transmitter set to model 1 when you were "going" to fly model number 2? Lose power to your transmitter for 2 years, or break your transmitter, no need to reset everything for 5 or twenty models.

Just basically ideas and a bit of hardware to play with for now, but there are plenty of great sources of info out there.

RC Pilot Project (http://www.uavs.net/)
www.steveonweb.com (http://www.steveonweb.com)
Paparazzi UAV project (http://www.nongnu.org/paparazzi/index.html)

I have the horizon sensing board available in kit form. Once I get more finalized for the main controller board, I may make that available too. The first generation horizon board kits have very small components and good eyesight and patience is required. Second generation horizon kits will have larger surface mount components.

www.scalerobotics.com/store/catalog (http://www.scalerobotics.com/store/catalog)

[LukeZ]

Well I don't have any pictures to show, but here's the details of my UAV. It's pretty much still in the early development stages.

The onboard computer is a PC104 format 586 (the MZ104 (http://www.tri-m.com/products/engineering/mz104.html) from Tri-M). I'm new to PC104 and I'm still learning a lot, but they're basically small computers, which I felt I could figure out better than microcontrollers. I'm starting to think that may not have been the case, but anyway I've paid my bucks for this thing so there's no going back now.

The MZ104 has two serial ports, a parallel port, and the usual keyboard and mouse. I have 64 MB of Ram and a 200 MB DiskOnChip solid-state hard drive, all in the size of something not much bigger than a deck of cards. Currently I'm running DOS on it and am still trying to figure out how to get it to load from ROM so it can boot quicker, no thanks to the tech-support people... It may be impossible. If so, and if the PC has to reboot in flight, it takes about 10 seconds to reload the OS from the hard drive and start running. So I'll need to figure something out to keep the plane stable if that occurs, but that shouldn't be too hard.

I don't yet have any sensors, GPS, or telemetry system set up. For the sensors I'm going to go inertial, probably using accelerometers and piezo gyros from Analog Devices. I'd be very interested in learning from anyone else who has done this. That's the primary reason why the Paparazzi project is of limited use to me, because they rely on horizon sensing. My UAV will be flying at night and perhaps in IFR conditions so that isn't an option. The Autopilot Project (http://autopilot.sourceforge.net) uses an inertial system but for helis, so I may end up having to modify their code, as it's all that I know is available. So far when I look at it it's been mostly greek. From other people who have already asked the question I know that it will need to be fairly extensively modified for fixed-wing. The Paparazzi folks also talk about doing inertial eventually but so far they haven't gotten around to it. I'd be happy to work with anybody here to develop code for an inertially-guided fixed wing platform, as that seems to be one of the holes in what is currently freely available to the hobbyist.

The goal for my UAV is for it to travel very long distances, on the order of 2,000+ miles. So telemetry will be a challenge. I will probably go with a cellular modem. I have my Ham license but the more I think about it the more I don't see how amateur radio would work over such distances. I will probably use something like the MaxStream for testing though.

Finally, did I mention I want it to be electric powered? :cool: I don't know if that design criteria will be possible or not. I'd even be willing to have a gas powered generator on-board to recharge the batteries... I haven't experimented with this yet, and am not sure how efficient it will be. Anyway, that's a goal much further on down the line.

My next step is to start writing all the code for this project. I'm a Visual Basic programmer in real life, though I have a feeling it's not really going to transfer over much. I've never written infinite loop programs or interfaced with the PC on a low level...

Like I said, I have lots of learning to do... but I'm in no hurry.

[radiohound]

Luke, great to hear about your project.

You are right, in thick fog or within a cloud, the IR sensors for horizon control would probably not work. However as long as the view toward the ground is warmer than the sky, it will work night or day. What I have noticed that clouds seem to usually measure around freezing. Clear conditions give negative 35 degrees f or lower near zenith, which is off my scale. From what I hear, this was first developed for the satellites. There is more info here http://www.ctie.monash.edu.au/hargrave/horizon_sensing_autopilot.pdf

So, in clear conditions, make the ground temp is above about -20 deg f. In cloudy or overcast conditions, make sure to stay below the clouds and the ground temp should be above freezing.

The big drawback are false horizons, like mountains on one side. This is where some other form of sensing would be a very good thing.

Walter

[Arp]

If interfaces and computing power are available to handle 2 webcams (4 would be better, but 2 seems to suffice) and run fairly simple image analysis..

...then during daytime, provided sufficient altitude... one could compute the horizon in *quite* robust fashion by detecting the radically different color/texture of land and sky. (Method has reduced viability above sea and snow, but a sufficiently smart algorithm should distinguish even those from sky.)

At sufficient altitudes, the resulting system would hardly err.

----------

What bothers me... is that I want a system which could reliably determine "up" and "down" indoors. That is, unfortunately... notably more difficult. Proper orientation appears to require combined senses, for example:

-- Accelerometers ("I feel acceleration in that direction. Consequently this direction should be down, or at least opposite to my direction of movement.")

-- Causal measurements ("When I reduce throttle, I should generally start falling down. I will now reduce throttle, and see which way I start falling.")

-- Image analysis and memory ("Last time I measured the direction of 'down', I determined that downward is located a grey surface, while upwards is located a white surface. Since I haven't moved a lot, this might still apply.")

-- Motion prediction ("Since I started moving left, I should not mistake the acceleration I feel towards right for gravity. This acceleration results from my movement. Until I have performed a freefall experiment, or fixed my position relative to a standing object, I should not trust my accelerometer to tell where 'down' is located.")

Such logic, especially where it must operate with high reliability. Thus, I try making utmost effort to equip my slowly-forming pet project with an airframe exhibiting some inherent hover stability, and ability to survive one or two temporary errors.

[LukeZ]

Radiohound, I've been thinking about the servo control on your setup. I agree that it would be pretty nice to have the setup stuff be in memory on the "receiver in the model." I take that to mean that the receiver in the model is actually just your PIC, taking serial commands from the radio link, and translating those to signals for the servos. You could save stuff about the servo reversing, mixes, etc in memory on the PIC.

That's cool, but are you going to have a manual control mode? If so, and if the servo control information is being sent as a set of serial commands through the Maxstream modems, what are you going to use for a controller? Joystick on your laptop? Or are you going to have a standard rc receiver on board, with some way to switch control from it to the PIC?

[radiohound]

Hey Luke,

from the datasheets, it looks like the Maxstream 9xtend radios are fast enough to perform transmitter, receiver, and telemetry functions. No traditional receiver necessary. I purchassed a used RC radio for $25, ripped all the electronics out, and plan to test my theory.

The pic18f876 I will use has five 10 bit a/d converters. I will connect the four stick potentiometers to 4 of these inputs. Trim buttons will replace the analog trims that are on this (previously analog) rc transmitter. Eventually, I will use the traditional gear switch to bring it to autopilot.

My first steps will be to build the transmitter receiver group using the Maxstreams. Once I get that running, I will work on the telemetry back down to the "transmitter".

The next step will be to work on the autopilot. This will be performed by two pic chips, a garmin Geko gps, 15 bit altimeter, air speed sensor, infa red sensors, and a pretty stable airframe.

A long term project, but lots of fun.

[mattweisz]

Radiohound, I also plan on using maxstreams to control a model aircraft. The plan right now is this: JR Transmitter->fms interface cable->pic18F452->9Xstream->======>9XStream->PIC18F452->6 servos.

After that project, I plan on buying an IMU and writing some code to control a helicopter or airplane. Should be a piece of cake...Just Kidding, It'll prolly keep me busy for quite some time.

matt

[radiohound]

Matt,

I am very interested in your project. Your 32k Pic processor has me thinking. I am curious though why you selected a 40 pin Pic? Now I am tempted by the 28 pin brother, the PIC18F252 with 32k! I would hope that I would never run out of code space with that one, and it has the same footprint as the PIC18F876.

Keep us posted on your progress! I like your trainer cable interface idea too.

[mattweisz]

The only reason I chose the 18F452 (TQFP version - it's tiny) was because I have about 20 of them just lying around...and I have some boards available for it The trainer cable interface is only in the concept stage now...not really sure how it'll work.

How do you like the 9XTends? We have some, and they get pretty toasty running at 1 watt.

[radiohound]

I haven't quite gotten to the stage of trying the 9xtends yet. I skipped the development kit, so my pic boards will be the development kit. I have a few more things to finish, trouble shoot, then I need to try to come up with some code to make them talk to eachother.

I hadn't thought much about the heat. I have some pretty short transmissions planned, with probably less than 25% transmit duty cycle, so hopefully it won't be a problem. Also plan to use rssi (received signal strength indicator) for automatic power selection. But just slightly into development stage, so I'll see what happens.

Walter

[joelhaasnoot]

This thread has died, but maybe there's people who would like to contribute now ;) Seeing what other people are doing is always great!

[funkysnail2k]

hallo,
I see this thread hasn't been replied to for quite a while, so i thought i would see if i could start it back up!

I was wondering how people's projects are getting on and any advice anyone may have.

For my post grad masters i am building a UAV, my supervisor's aim is to have it automatically land, mine's a little simpler - stay up and possibly stable for 5 minutes. as yet i haven't got very far, my airframe is a ripmax harmony, and for the control board i have used a DS30 Pic with several sparkfun acceleromiters and gyroscopes, the pic has several break out pins for communications to extra perhiferals etc, including a serial port to more powerful computer (if needed).

I am looking at state space modelling to create a decent simulation of the plane, from this i can devise a controller where i can specify angles of atittude or velocities (rates of turn) and it does it.

At the moment, the pic board needs to record everything from attitude, altitude, airspeed to servo positions, to gps location and transmits all the data to a down link - one problem (of many) i'm facing, can anyone recomend a transmitter i could use for a down link?