Uav

[trash]

My local radio club have set ourselves a small task of flying a UAV.
This isn't the typical Radio Control or powered UAV of which there is already a reasonable amount
hardware available and writing code for it is simple.

This task is basically a glider, and the aircraft needs to be fully autonomous.
If you think of it being thrown out the back of an aircraft with some skydivers, that's the project.

The UAV firstly will free fall and be weightless until there is enough air flowing over the control surfaces for it to be able to steer and right itself.

Once it has done this it will then work out where it is by GPS and of course it will know where it has to land. It will have no knowledge of local wind conditions.

This task is very easy for a sky diver, even though he's not working on GPS, he can observe and react to his surroundings.

The type of aircraft has not been selected yet. It needs to be able to survive being thrown out the back of a plane, so a lifting body is one of our choices. It will also offer a little bit of protection to the payload should it smack into the ground a little hard.


The typical flight would be a weightless start with air moving over the aircraft to naturally stabalise and right it as if it were recovering from a stall and the wing begins to fly.

Next is to pick a heading to the landing area. That's simple too.
Keeping the wings level and the aircraft flying straight is also simple.

Now comes the tricky part.
The problem is that in our club were have several pilots and avionics people and we all have our set ways of flying. I think this corrupts our thinking a little so it is the reason why I have brought the subject here to get some fresh thought on subject.

---

We have two modes of flight for the landing. The first is simple. Just fly the aircraft to the landing zone and fly circuits in a level long glide to the ground. It's the slow way down, but it is reasonably easy to write code for. There are reasons for not taking this flight path. If the aircraft has a good glide ratio it may be thermalling for a long time.

The second flight mode is the preferred but is much more tricky.
It follows a typical sky diver's flight or a helicopter in autogyro or a space shuttle landing.

The flight is a nose down and run for the landing zone.
The aircraft picks up a lot of speed which gives it good control and enough energy to make it to the field.

The thought is how to control the aircraft's pitch to hold the nose down in the dive and work out if the dive angle is correct. In human terms we think of flying through a corridor to the edge of the landing field.

The next question is how to work out where the ground is.
The skydiver and helicopter see the ground and they flare their landing before they hit it.
The space shuttle does a similar thing too. Though like our aircraft it is expected to have a high velocity on the decent and it may need make some turns to wash off speed before the landing.

And example is it might make a couple of turns in the last 1000ft.
In all cases we might assume that we have made it to the downwind approach with 500ft to spare 45% glide slop and enough airspeed not to stall.

The last part is the hard part.
How to get the aircraft to flare on landing. This might be a bit much to ask.
It will not have any landing gear. It will be a controlled crash. The trick is how to minimise that crash to allow the camera and electronics to survive.


There other problem is that if for some reason the aircraft gets lost, as it comes close to the ground in any case it still needs to be able to land itself.

The main problem boils down to needing to calculate or detect how far away the ground is and how to do it quickly in real time knowing that the ground may be any terrain and not knowing before hand what the local QNH or ground level will be.

How not to hit the ground hard (without a parachute).

 

[misterT]

Can you have electronics on ground at the landing site? What is your intended budget?

 

[jpanhalt]

The flight is a nose down and run for the landing zone.
The aircraft picks up a lot of speed which gives it good control and enough energy to make it to the field.

Have you considered a third flight path option?

Option 3: Nose up to give a very steep glide, then level at decision height and land as normal. Clint Eastwood did that in Space Cowboys, but it can also be used in real life. You could have a stall detector on the wing to take care of the change in stall speed with altitude.

How high are you planning to launch from?

John

 

[trash]

The first couple of tests are off a 2000ft hill. The first will be under RC to make sure the aircraft can actually fly.
The design of the aircraft is still being refined. For the time being you can assume it's a lifting body of some sort.

The first autonomous tests will be from the same hill. The landing zone well within even a lousy glide ratio.

The first real free flights will be from 12,000ft. It will just be carried out with one of the sky divers who'll let it go when they're clear of the aircraft if the wind doesn't rip it from his hand first. They'll free fall away so there's no chance there will be any sort of mid air collision.

Did you have some sensor in mind as a stall detector ?

One kind I've seen is a simple air speed switch. A small spring loaded tag on the top of the leading edge of the wing. When air speed over the wing drops, the switch activates. It's simple and can be made small and light enough for a small glider. Air speed drops, elevators down. Airspeed too fast, pull up.

That may be a solution; keeping a constant airspeed will cause a predictable glide. (with respect to airspeed)


Option 3. Did you mean nose down for a steep glide ? That was the second option. To dive at the landing field and flare to wash off the speed.
The concern is when to pull up. Too soon and you over fly the landing zone. We don't want the MCU to decide it needs to make a 180 degree turn at this point Though large circles might be an option. I'll have to test crash that one to see what vertical and horizontal speeds might occur.
Turns cause a decrease in ground speed, but an increase in vertical descent rate.
This kind of crash might be an acceptable method of landing.

I should mention that the aircraft will probably be styrofoam, so it is designed to withstand a reasonable crash and be easily repaired or replaced.
Most of the electronics will handle a crash well or are easily repaired. Consider something like a Go-Pro camera on board which we don't want to break.
And given that it will be in the nose of the aircraft, it's more of a consideration that we don't want to slam into the ground nose first.

A consideration was made as to even deliberately crashing it into trees. Towing an arrestor string. The aircraft flies into the trees with provide the braking and the arrestor string hopefully catches in a fork of the tree before it hits the ground. It was a whacky idea, and we might even test it. The issue is that the tow line itself is proportionally heavy to a small aircraft and may cause it to stall in flight.

There is no idea that we will not entertain. The ultimate goal is to have a lot of fun and get some good video back from the project.

 

[Nigel Goodwin]

The first couple of tests are off a 2000ft hill. The first will be under RC to make sure the aircraft can actually fly.
The design of the aircraft is still being refined. For the time being you can assume it's a lifting body of some sort.

The first autonomous tests will be from the same hill. The landing zone well within even a lousy glide ratio.

I would suggest the autonomous tests should be a long way off, as you say first make sure it flys (well!) under radio control. Second, take it up to the planned launch height (probably many, many times), and try flying it down on radio control - this should identify the problems.

Then start adding the autonomous part, from a low height and work up.

 

[jpanhalt]

Stall switch is as you describe. It is typically located about mid way out the span and senses separation. That is, you want to sense the stall before you lose aileron control.

By option 3, I meant the nose up attitude to give an airspeed less than max glide, but slightly higher than full stall. Putting the nose down to land short is often a poor choice because of difficulty with speed control and glide angle. Go to a glider airport and see how they do it, without spoilers (if you can find one), but spoilers do a similar thing. You want to kill lift, not point like an arrow for the ground.

Some aircraft can be designed to actually do a falling leaf approach (deep stall controlled mostly with rudder, not ailerons), followed by recovery. Early aircraft (e.g., WWI) used to do a spin to get through a cloud bank. A true spin is a stalled condition that loses altitude fast without resulting increases in airspeed or stresses. I would not recommend either a deep stall or spin for a pilotless glider, however.

John

Edit: You might also consider "dethermalizing" , which is a stalled condition used by model free-flighters and SpaceShipOne. It would probably be more stable than the falling leaf or spin mentioned above.

 

[trash]

Stall switch is as you describe. It is typically located about mid way out the span and senses separation. That is, you want to sense the stall before you lose aileron control.

So the question is how to build something suitable for a foamy for testing.
Pretty simple thing for a PIC chip to do with servo elevator control flying a straight in approach.

Putting the nose down to land short is often a poor choice because of difficulty with speed control and glide angle.

Not in this case it is a useful attribute of the flight and landing control.
I do fly a couple of different types of aircraft and this is not an unfamiliar way to fly.
No I don't have passengers

Some aircraft can be designed to actually do a falling leaf approach

That's worth considering, but I guessing such aircraft might not fly well at high airspeeds.

I would not recommend either a deep stall or spin for a pilotless glider,

I don't recommend them (spin) even when I'm the pilot
A spin is definitely not a desired condition for this aircraft.
A stall however is a useful manoeuvre for losing airspeed and altitude quickly.
Like the above leaf landing, stalling the aircraft 100ft off the ground and slowing it's airspeed for a controlled crash in the right spot could actually be an acceptable method for getting the aircraft back. Obviously the destruction of the aircraft is a consideration then.

You might also consider "dethermalizing"

I've never heard this term before and a quick look at the description and I understand what you mean. It's actually a technique I already use paragliding (big ears). https://upload.wikimedia.org/wikipedia/commons/f/f1/Paragliding_big_ears.gif
The Wikitionary description made me laugh.

The method that space ship one uses is already a consideration for the first part of the flight while it's in free fall. I haven't considered (yet) if it has a use for landing.
I'll be pondering that option for a while.


I'm no stranger to flying, aerodynamics, RC aircraft, electronics and programming.
Actually making an MCU fly the aircraft in free space appears straight forward and getting it to the landing area is also relatively straight forward.

The desired goal is to get the aircraft to land in a paddock without a parachute etc and without damaging the payload. To have it land in a preferred paddock, but if I can't make it back to the landing field, then it needs some way of getting itself safely to the ground in an unfamiliar environment.

I think about how I approach an unfamiliar landing area in a glider and try and apply it to an MCU. The problem seems to be I have a sense of where the ground is.

The problem makes more sense if I think "How would I land this aircraft blindfolded?"

Ground proximity radar would be a great help