I've been working for the last few weeks on developing a parachute recovery system for E382s to reduce the need for manual flying. I know from my own experience that any time a human is at the controls, the risk of crashing is significantly higher. Parachutes have been made for small aircraft before but they're often implemented in an expensive or complicated way. The idea here is to make a simple, working parachute that reduces the overall risk of damaging the aircraft, is simple to use and cheaper to install.
Building the parachute itself was pretty easy, I used this gore size calculator by Scott Bryce and just printed the gores out as patterns. I accidentally printed the pattern out on 8.5x11 instead of 11x17 so this parachute is actually only 22" diameter - much smaller than what it should be for this weight. The gores are cut out of black rip stop nylon and just sewn together. The lines are attached at 8 points, with the lines simply tied through button holes. The line I used is too thin to be sewn into the material but I prefer the thin line, made of kevlar, for its weight and size.
Next I ran a few tests using some dummy mass at about the same weight as an E382, 1.70 kg, to see how quickly it would fall and if it would even open up properly. The first few tests were done with the parachute already pretty much deployed as it was from a pretty low altitude off my building's fire escape.
I thought the speed it hit at was slow enough, despite the smaller chute size, to not significantly damage an airframe and decided to move on with the tests. So I attached the parachute to a light-weight (1.5kg) airframe to see how it would fall and if the wings would affect the speed significantly. Because of the smaller size of the parachute, the airframe was actually still had almost enough lift to fly itself. So as it fell and built up speed, it would begin to pull out of the dive, then the parachute would slow its horizontal movement and it would stall, then presumably repeat until it hit the ground. In a few of the tests, the parachute line also got caught under the horizontal stabilizer, further complicating the problem.
Not wanting to call it a day, I pressed on to simulate the actual deployment. Ideally, the parachute would deploy on its own, be near the center of gravity and allow the plane to fall down roughly level and impact the ground on its belly. That all points to a deployment from on top of the wings, right in front of the motor. On the first try, I loosely packed the parachute on top of the wing and just tossed it. This allowed it to catch wind almost immediately which promptly pushed it straight back into the tail, where it lodged itself, allowing the plane to crash at high speed.
But I tried again, packing the parachute more tightly so it would roll out past the tail in a compact ball before opening up and catching wind. That did the trick! The dive/stall issue is still there but that should be fixed with the next parachute build.
All four tests above plus one more live deployment test on a flying plane are in the video below.
Would you ever use a parachute as your primary recovery mechanism? Why or why not?
Comments
@Randy
Thanks a bunch. Just what I needed.
Rob
Hi Randy, If you get a cheap light flyrod, you could cut off the tip of the rod and bend it back on itself, then all you need is to design a little servo trigger release mechanism, not only get the chute out of the bag, but deploy it past the props as well.
I proposed one several months ago that used 4 rods to deploy a square chute where the corners of the chute would be directly attached to the ends of the rods, so it couldn't get tangled and given a few meter drop would even be self righting.
However since then my copter has been so reliable it hasn't really been a necessity (talk about famous last words).
@Rob,
Re the deployment...or maybe the bottom level could be on a hinge so this whole level on the bottom of the copter opens like a hatch door. It would have the battery on it so it's very likely to open quickly. Might be a bit messy if the hinge isn't really strong or the battery isn't attached securely.
Re detecting if things have gone wrong...I'd set it up as another failsafe so I'd put it in the events.pde along with the battery failsafe, gps failsafe, etc. Basically I think we need to check for a persistent difference between the desired angle and the actual angle. The stabilize controller is the most commonly used controller (it's used both in stabilize mode but also higher autopilot modes also use it)...so you could modify get_stabilize_roll and get_stabilize_pitch so that they record the average absolute angular error for the past few seconds (maybe put it through a low pass filter or something). Then you could add another function that runs at 1hz (i.e. make it be called from the slow_loop()) and checks that average. If it's too big you deploy the parachute. For deploying the parachute...well...the camera library shows how to move a servo although there are various other bits in the code as well that do that.
Randy,
I like the whip idea. Simple, fast, and won't light the forest on fire when it deploys. Con, a weird whip thing sticking out the side. But we already seem to have a bunch of stuff protruding from out flying things, why not a whip?
I've been thinking of a Pyrodex charge, or just popping a latch and letting a chute fall out. Your idea is interesting.
I've been contemplating forking the code and experimenting with activating a servo when "the controller thinks things are going horribly wrong". Do you have any guidance for an APM noob as to how you would approach that? I am just starting to get familiar with the code, and your opinion on how one would implement such a feature would be much appreciated.
Thanks & regards,
Rob
I know this is a slightly old blog post but if people are following...
As a "background task" I'm thinking about a parachute deployment system for my copter and I have purchased the one below from opale models.
I've been thinking that there needs to be some kind of long-ish whip like pole which can yank the parachute out of the bag. Do people think a plane landing gear servo might work? like this for example? hobby king has a whole shed load of various types of retractable landing gear...
@Marcus
Didn't mean to give the impression that I am opposed to a recovery system. Far from it, I am working on one myself. The only thing holding me back in my droning joy is the fear that my flying thing will crash through somebody's windshield or on top of their head.
Once I have a working ballistic chute, look out world.
@ Rob Bartlet
We've had copters fall from 100 m for no apparent reason with $2k worth of cameras. You can't always fly them out of trouble. We need a light, robust, configurable, and autonomous recovery system.
something i was exploring was a gravity induced opening mechanism, I liken it to a trap door, fixed to the cg on the aircraft. also in model rocketry books they have formulas for figuring out what size parachute you need.
Hi dears...
Bought 16g CO2 cylinders and try testing with solenoid valve for ejection chute. Pyrotechnics seems to be interesting for lightweight aggregate. I'll be here watching the topic.
@keeyen
I don't know about that. It seems to me that at the minimum you should be able to exercise full control over the basic flight operations of the aircraft-- take off, landing, level flight, etc. It seems to me that flying a drone is more than just executing a mission. You need to be able to deal with unexpected glitches. Popping a chute at the first sign of trouble, or even when there is no trouble, seems to me makes a very poor pilot who arguably shouldn't even be in the air.
But then I can see the argument that a chute is just another flight mode. Still, you should be able to FLY the thing no matter what, IMHO.
-
1
-
2
-
3
-
4
of 4 Next