Questions regarding UAV as a design project

[flatland2d]

Greetings all! I am working with a group of 3 other electrical engineering students in a senior level design course. We all have great interest and experience in robotics and most of our ideas so far revolved around that. Then we got to thinking how impressive a UAV would be to build. For the course, we will spend this semester writing a design proposal (and getting a head start on the actual design) and need to have the design done (or at least something to show for, explain why it didn't work) by the end of the spring '07 semester. However, I am the only person in the group with R/C airplane experience, and I don't have a whole lot at that. I can fly a plane steady and have a pretty good understanding of the controls and aerodynamics however. Unfortunately, we are very funds limited, so this design can probably not exceed $1000 by too much. That may be somewhat flexible, but I'm not sure exactly how much my other group members would feel comfortable contributing.

Ideally, our UAV would be equipted with GPS and a wireless camera and be capable of flying to a prespecified location and circling around. As an extension to that, we would like the UAV to be able to fly to GPS coordinates being broadcasted by the transmitter. That way you wouldn't have to preprogram it to fly somewhere, it would just listen to the radio transmission sending out its GPS coordinates. Theoretically the radio transmitter could be mobile (such as placed inside a car) and the UAV would track the car. Landings would be done under human control.

We are aware of several flight control modules out on the market. There are also some other students doing research at our university on a helicopter UAV using one of these systems. Due to their high cost, we are doubtful we would be able to afford such a unit given our budget. Being the "hands on" people that we are, we are considering making our own control module using accelerometers such as those listed here (http://www.sparkfun.com/commerce/categories.php?cPath=23_80). Would this be a feasible thing to do, or are the control problems so complex that we wouldn't stand much chance finishing the project on time? It sounds easy enough to continuously poll the sensors and make adjustments, but I'm sure there's a reason why the modules already out there cost $1000! What problems might we face in designing one ourselves? Being that the accelerometers are a MEMS device, how would the mechanical vibrations factor into this? Would we see tons of noise?

Now, as for the specifications of the aircraft:
1) Fixed wing aircraft, preferably of high wing design to go for something that is more naturally stable in flight and easier to control. Something on the "trainer" level.
2) Enough spare power to haul around a few extra circuit boards and associated electronics. We would also like to implement a wireless camera, like those sold by Black Widow.
3) Sufficient runtime to fly to a location (possibly no more than a mile away for proof of concept) and circle around for a few minutes, then return. We will probably want at least 15 minutes as a bare minimum.
4) Easy to repair. Even though we will always plan on having a human ready to take control, some crashes are bound to happen.

The underlying question in all that is gas vs. electric. I'm used to foam electrics and they fit all the requirements except runtime is pushing it. For that reason alone we may want to go with a gas plane (and much increased power). However, all the gas planes I've seen are wood, and they seem a bit intimidating to repair. Do they even make foam gas planes, or would the heat melt them in no time?

Thanks in advance for any help. Right now we are still throwing ideas around and trying to decide if such a project would be feasible. Any input whatsoever would be much appreciated.

[stone_axe]

Foam glow planes have been around long before modern electrics. There is no reason not to use foam if you want to use glow. Using actual gasoline engines presents a problem because gasoline and it's exhaust melts most foam.

The main problem with using glow for a cheap UAV is the issue of stalling the engine. Most commercially available glow engines make great power at full throttle, but are not optimized for lower throttle settings. There is a significant risk of stalling a glow engine when flying at low throttle settings for extended periods of time - and especially when going to full throttle after extended low throttle periods. There are precautions you can take to avoid this problem, but it is still an inherent problem.

Given the fact that losing the motor on a UAV means a mission failure, that risk is too great.

Modern electrics are actually more powerful than glow, especially in short burts. An electric drivetrain can deliver much more power to weight by overloading it briefly as needed and then can be throttled back to delivery maximum efficiency at cruise. The idea would be to optimize the drivetrain for the cruise profile, not the burst power profile. This where electics excel, because you can fine tune your motor and prop selection to match your optimal speed and thrust for the cruise profile. Also, with brushless electric motors and electronic speed controls you can achieve the same efficiency at cruising throttle as you can at full throttle.

If you use Lithium Polymer batteries and brushless motors, 15 minutes of run time is the norm. You could easily get 30 to 45 minute run time if you simply used larger LiPo batteries.

[fhhuber506771]

visit www.spadtothebone.com

You could go to a variation on that theme... corrogated plastic signboard construction. If you use 2 mm for wing skins you can easilly keep weight under control. Some of the coroplast models come out lighter than the balsa based construction equivilents.

The material is relatively inexpensive. Harbor Sales will ship you the coroplast if you can't get a sign shop to get the 2 mm stuff. (4 mm is easy to get) The main problem with mail ordering the material is that you either pay a UPS surcharge for oversize to get 4ft X 8 ft (or 4 ft X 12 ft..) sheet... or you get it in 2 ft X 4 ft.

assembly is quick once you have the plans also. A complete .40 size RC combat model can be assembled in 4 hours including radio installation.

Gas vs electric... well... its not difficult any more to get decent durration with the electric. LiPo batteries make it work. The problem with That route is cost of the batteries. They'd eat up your whole budget.

If the project is more about the robotic control system then designing a "better airplane" then I would base the model design on the "SPADET" or "Debonair" style... basic rectangular wing typical of an RC trainer, very conventional appearance. (mimmicing the appearance of a Cessna 152 essentialy) Just.. bigger for the payload. the Debonair is "overbuilt" ti allow for abuse given a typical trainer and can easilly have weight reduced, allowing larger payload.

[stone_axe]

If I was designing a UAV I would not use a conventional rectangular winged RC trainer.

There may be some advantages to using this tried and true planform, but it is not very efficient.

I'd look at sailplane and cross country lightplane designs. I'd optimize the dsigne for the cruise profile and not the takeoff and landing profile. I'd use flaps and the ailerons to handle takeoffs and landings.

[flatland2d]

So I'm guessing it wouldn't be possible to use a glow engine for the reasons stone axe suggested. Are there no glow engines made to operate at low rpm? We can sacrifice maximum speed for a slower running motor that is stable if there are any available. This is ultimately proof of concept so 200mph speeds are definitely not required, and would probably be avoided for safety reasons!

I have not come across any foam glow planes, so if you could suggest some, that would be great!

I would actually prefer to use brushless motors over glow engines, but I didn't think they'd be up for it. I'm glad to see they are still a possibility. However, the battery issue could be a problem. Say if we went with a brushless motor sized for a plane with a 72" wingspan (not very detailed, I know, but bear with me, just trying to define the approximate size we are after), how many mAh would be required for those runtimes? What might we expect from four 11.1V 1250mAh (5000mAh total) lipo's in parallel? That'd cost about $200, which we could probably afford. Theoretically that'd support a 10A load for 30 minutes, which would be nice. I'm guessing appropriately sized NiMH cells are out of the question? Is there any way to make them work (design the plane to carry heavier payloads)?

fhhuber, thanks for the SPAD link. We would like the idea of being able to custom design our plane. Until now, we were thinking of using a kit and modifying it for our purposes.

We are unconcerned at this point with developing a better plane. We just need something that is naturally easy to control to simplfy the control equations of the plane.

Any thoughts on the control aspect of the project?

[Tom Harper]

I have found Foam Board to be the most cost effective material. It is easy to cut and bonds with carpenters glue. I've used it on models as large as 14 sq ft. It is especially easy to repair. I completed one in 20 hours of building time.

An 8 sq ft version of this model ( www.sdc.org/~crane ) would fit your need. I have used one for a couple of years with a Norvel .40. The take off speed is 15 to 20 mph and it will cruise at 25 mph. You don't need to idle the engine in the air. With a large airplane you can cut to half throttle and fly with good engine control.

Electric vs Glo - depends on the application. I use glo because it's cheap and durable. I can go to a remote area and put in four sequential flights of 15 minutes duration each in quick order. I can't afford four or more Lipo packs and I'm not going to wait for the batteries to charge between flights. On the other hand it would sure be nice to step out of the car, flip a switch and be in the air.

[kd7ost]

I think your deadline, and the fact that only you have RC experience in your group severely limits your project. If you're going to be building such an advanced guidance package, I would simply buy an ARF with a glow engine. A good quality glow engine, broken in correctly, and adjusted correctly should be able to idle without loading up and quiting. We do that all the time. Idle is for sitting on the ground and landing approaches.

The term UAV here doesn't mean it has to stay aloft for hours and hours. And you don't have time to build a plane from scratch. I would look at the Hobby Lobby 8 foot telemaster, or equivelent airframe. Get it flying as a test bed and work on the rest of the advanced parts. Don't re-invent teh wheel, or plane as it were. You have your hands full as it is.

Dan

[Unterhausen]

You don't have that much time. I would seriously look at the Sparkfun autopilot. I would recommend going with electric, trainer, arf.

I don't think accelerometers are sufficient to fly. It starts to become a difficult problem with minimalist sensing. You probably need rate gyros and possibly magnetometers as a minimum. The other alternative is using thermopiles like the papparazzi project did.

[fhhuber506771]

If you are looking to fly an appx 7 to 9 lb model, roughly .60 to .72 glow power equivilent size using electric power:

An AXi 4120/18 motor turning a 13X6.5 prop would need appx 7000 mah worth of 6S LiPos for 25 min flying time.

Thats NOT all at full throttle...

The LiPo batteries would be a significant part of the total model weight. NiCd or NiMh, trying to get over 20 min flying time would be too heavy.

**********

The 4-stroke glow engines do better about extended mid-range operation than the 2-strokes. They also extract more useful power per ounce of fuel used. So glow is not really excluded from being practical. The Magnum brand glow power 4-strokes are reasonably priced and reliable. I think you get more power for the same total engine weight (including muffler) by going with a 4-stroke. The huge 2-stroke muffler vs the small 4-stroke muffler is significant.

*********

Time involved in designing and developing the ariframe would take time from your developing the electronics. Sticking with the tried and true "trainer" style design means just scaling the model rather than needing a lot of testing and experimenting to get a stable aircraft. Start with a plane that works well as a trainer, and you reduce the "workload" of your autopilot in keeping it on course.

BY going with a simple design and a quick construction method you can devote more time to getting the electronics systems working correctly.

Thus... I would choose the coroplast construction and a trainer design. You could have one of the coroplast trainers (scaled up if needed for your payload) flying in under a week with ease.

[Tom Harper]

Gotta second the idea of buying a big ARF. The quality is excellent and the price is right. The emphasis on your project is electronics not the airplane.

[flatland2d]

A big ARF is pretty much what we're after and the Telemaster fits the bill. At $90 it's possible we could just buy another one if we had one or two disatrous crashes (though there are cheaper kits out there, any recommended?). I do like the coroplast and foam designs, but I think, as others have suggested, it would take too much time away from developing the electronics.

Thanks Unterhausen for the info about sensors. This what I really need to know. I will look into rate gyros and magnetometers.

I'll also check into the possibility of using a 4 stroke engine. As for electric power, I'm kind of thinking it'll simply cost too much, AND the downtime required to recharge the batteries for subsequent tests.

[Unterhausen]

$90 is for a box of die cut balsa, not an ARF. Probably not that hard to put together.

I saw the kit for the 12' Telemaster, it was in a very small box. I was suprised by that.

[Ailuropoda]

I'd like to suggest considering weight and run time to be a side issue. If the electronics wind up squeezing everything else out of the airframe, finding someone with a larger model to tow it to altitude and having it glide some kind of patern and do a controlled landing proves that it works. It doesn't make a very impressive demonstration, but it demonstrates that the electronics work correctly.

[Unterhausen]

I'd like to suggest considering weight and run time to be a side issue. If the electronics wind up squeezing everything else out of the airframe, finding someone with a larger model to tow it to altitude and having it glide some kind of patern and do a controlled landing proves that it works. It doesn't make a very impressive demonstration, but it demonstrates that the electronics work correctly.
I feel the same way about flight time. I don't think the electronics will weigh that much, but if you can't afford bigger batteries, go fly and tell the people evaluating you that if they give you an extra $200 it will fly twice as long. Works for me, and I'm an academic.

[flatland2d]

Yeah, runtime isn't such a big issue. The concept would work with even just 10 to 15 minutes assuming the plane just went there and back without circling around for too long. 20+ minutes would just be a nice feature for extended tests.

Electronics won't weigh much at all. Weight really isn't an issue as I see it (assuming a glow engine plane). Just a few circuit boards, small sensors, GPS module the size of a quarter, Black Widow AV camera and transmitter, and a small battery. We have thought about mounting the camera on a turret that could be human controlled from a second transmitter.

I've scratch built model wooden boats so I'm pretty sure I know what to expect putting together a kit with parts already cut out. ARF would just be nice to save some time.

I think I'm just going to have to test the theory of operation and sensors to see what is feasible. I still have my old foam plane and I was thinking of taking the wings and mounting a dowel perpendicular to the wings in place of the fuselage. The dowel would be supported at each end allowing the wings to roll. Then I would place a high speed fan to simulate the plane flying and see if the control system would keep the wings level. A second fan could be used to simulate a cross wind. Theorectically the pitch axis would work in the same way. Now that I'm thinking about it, we could do a lot of testing in a home built wind tunnel. That'd help tremendously to prevent hard crashes.

Thanks for all the advice. Keep it coming!

[kd7ost]

Now that I'm thinking about it, we could do a lot of testing in a home built wind tunnel. That'd help tremendously to prevent hard crashes.


That just cut your building time down a lot and soaked up some funds.

With an 8 foot Telemaster, a 20 oz fuel tank and a .71 to .91 four stroke engine, you could easily get flight times of 45 minutes. That's a little conservative and is provided you throttle back.

If you're shaky on flying go to the local hobby shop and find out where the local RC clubs are. Get involved and get someone to help you. You need to shoot touch and go's as well as get the plane set up for sure, but gentle flight. Don't try to work with radical throws.

This sounded like an order but it's not. It's just in my own experience what would get you moving along the fastest. You need to be very comfortable with getting the plane in the air and back on the ground. It is your vehicle and you have to handle the manual flight parts with confidence.

Dan

[flatland2d]

Thanks for the advice kd7ost. Maybe I was a little too ambitious sounding about the wind tunnel. I wasn't really thinking of anything more than a high speed fan (already have) and a rig to support the plane. The idea is to simply test how well the control system can fly the plane level and respond to small impulses on the wings. I want to know that when I flip the switch to go into autopilot mode that the plane isn't going to do anything disastrous. That's about all I'd use the wind tunnel for.

I'll be sure to seek out some local help before we get the plane in the air. I'm familiar with the setup procedures, but it'd be nice to have a trained pilot ready to take over. Believe me, we won't be experimenting with any radical throws. I guess you could say we plan on flying this thing "to scale", ie. not banking the plane 90 degrees, or rolls/loops for that matter. We want something as docile and easy to control as possible.

I might even see if I can pick up a copy of RealFlight on eBay or something. I've been wanting to buy an electric heli, and that'll sure pay off when the time comes to get one, though I don't know when that'll be.

I think I just need to order some accelerometers and gyros and start playing with them. I've got quite a bit of experience working with PICs, so that'll probably be our processor of choice. I'm eager to try this self-stablizing wing idea anyway.

[fhhuber506771]

Open jet wind tunnels (essentiall a fan and some vanes to straighten out the turbulance) can be helpful for low speed flight analysis.... so yo are not totally out in left field.

You want a pilot to assist.... well San Antonio is a bit far for me for more than a trip or two per year. But there are a couple of AMA affiliated clubs in Austin (half the distance...) and I'm fairly certain there will be at least one in the San Antonio area. (I'd bet close to 5 clubs within 30 miles of you... without even looking it up in the AMA website.)

Sure... it could mean joining AMa and the club to get some training... but it may be well worth it in terms of survival of your project aircraft.

[Ailuropoda]

A wind tunnel is definately nice. I suppose it wouldn't be too hard to build one that's good enough for testing a very small model. Use your fan to suck the air rather than blowing. You'll get less turbulance. A good tube to use as a test section and some sort of gently curved inlet in front. Some kind of screen door for more turbulance reduction. You'll need some space around the model to avoid wall effects or sheer layer effects if you want to go with an open jet tunnel. If you're going to build an 8ft model, this gets increadibly big, specially if you want to put the real thing in the wind tunnel. Now that I'm thinking about this, I'm starting to wonder how big my ceiling fan is.
There's other ways to get the same effect which might work better for you. For instance, I know of one company that developed a UAV and did their aerodynamics testing by rigging up something on a pick up truck. There's some inaccuracies doing it that way, but it got them off the ground.

[flatland2d]

That's a good idea about using a truck. I hadn't considered the size/number of fans to lift an 8ft model. I was still thinking of my little foamy for some reason.

I'll look into finding some local AMA clubs. I'm aware of one already that used to have an airfield pretty close by, but that field got shut down and their new one is pretty far from me.