Check out those flips! From Hackaday:
Straight from the Aerospace Controls Laboratory comes a variable-pitch quadrocopterdesigned by [Mark Cutler] and [Jonathan P. Howe]. While real, full-sized helicopters always have variable pitch rotors, changing the pitch of the blades on remote control aircraft is a fairly uncommon modification. When it’s done right, though, being able to easily change the thrust direction of a propeller leads to very cool flights, like having an airplane hover nose down.
[Mark] and [Jonathan] identified two interesting techniques that a variable pitch quadrotor can bring to the table. The first is trajectory generation - because of the added maneuverability, their quadrotor can perform more aggressive banking turns when following a preprogrammed path. The second benefit to their design is quick deceleration. In the first video after the break, you can compare the deceleration rates of a variable pitch and fixed pitch quadrocopter. While the fixed pitch quad continues climbing after being commanded to stop, the quadrocopter outfitted with variable pitch rotors can stop on a dime.
Comments
Many thanks Mark for the prompt response. It's great to know that I'm using the right platform and am certainly appreciative of Bill Premerlani and the rest of the MatrixPilot open SW team's very capable support and genius behind the platform.
@Daniel: We used the Unmanned Innovation boards during the prototyping phase. Since then we have been using the UDB4 with a 2000deg/s rate gyro (see Buddy's comment above). Good luck with the variable-pitch quad!
It seems that in the test flights, the boards used seem to be Unmanned Innovation's osNanoQuad autopilot board as shown in this video http://youtu.be/l8LBEpjb58c as well as in p14 of http://acl.mit.edu/papers/GNC11_Cutler_uber.pdf. I currently use UDB4 and am building a variable pitch quad, and am just wondering if UDB4 can perform as well and what programming it'll take to get it near to this level of performance?
We haven't done that test exactly, but I'm fairly confident that the efficiency is quite lower with the variable-pitch setup, most likely due to the fact that we are using a symmetric airfoil with the variable-pitch blades instead of typical cambered, fixed-pitch blades..
Have you guys done any work to compare, all-else-being-equal, the thrust production efficiency (thrust/watt) of different propeller types? ie: typical fixed-pitch props, the props that you are using, or helicopter-style rotor blades?
I've seen a couple variable pitch quads, and they tend to use say, a 600-class helicopter tail gearbox (variable pitch, but no swashplate) with 450-size main heli blades stuff into them. I have wondered about the efficiency of heli blades compared to a more profiled blade such as a fixed-pitch.
Hi all, I work in Mark's lab and like him I'm a huge fan of DIY drones so I figured I'd reply with some more detailed responses to your posts. Glad to see so much interest in the variable-pitch quad and hopefully we'll see some DIY versions soon =)
@John Bond: Yes, all standard pod-and-boom outdoor RC helis are variable pitch, but they have a swashplate which is mechanically complex and fragile. The reason we use fixed pitch quads in the research community is the physical robustness of having only four moving parts (when they break it is our job as grad students to fix them). However, you won't find many (if any) variable pitch *quadrotors*. We designed this vehicle to combine the maneuverability advantages of variable pitch with the robustness advantages of not having to use a swashplate. The Axi variable pitch system we're using adds some complexity, but Mark has found it to be surprisingly robust in the lab.
@Bill Premerlani: First of all, your UAV dev boards are awesome. We use them all the time in the lab for our quads. If you're ever in Boston let me know and I'll give you a tour. What you don't see in the video is an additional ITG-3200 gyro (using the sparkfun breakout board) hot glued to the bottom of the UDB. The flips Mark is doing are closer to 1500deg/sec so the IDG500's on the UDB are saturate quickly.
@Bill Premerlani, @Robert Lefebvre: Bill it looks like you've done some great analysis on fixed- vs. variable pitch. We've done some similar stuff, and Mark published a paper comparing the two which explains the models that we used and shows results for things like efficiency, hardware time delays, etc. If you're interested you can find it at: http://acl.mit.edu/papers/GNC11_Cutler_uber.pdf
@Todd Hill, @Eagle: We wrote the PIC software from scratch, but you are correct in that the core functional components are not a giant departure from standard quad control code. At the end of the day your control law will spit out roll, pitch, yaw, and collective commands which can either be actuated by the ESCs (motor RPM) or the servos (variable pitch). Our software can use either of these methods individually, or a mode that mixes the control to both methods. In the video in this post, however, variable pitch is doing all of the control with a constant command (plus some feed-forward to minimize large RPM changes) going to the ESCs. Another thing worth mentioning is that Mark had to modify the ESC code so that the motors did not overspeed too crazily when you take pitch out. In earlier versions of our quad this would literally shake the frame apart =)
@Robert Lefebvre: You bring up a really interesting point about using variable pitch for large changes and RPM for small changes. We did find that using RPM (fixed-pitch) for control produced empirically "smoother looking" flights. This could be a result of the added motor lag in RPM mode to smooth the control response, the PWM resolution of the servos, or the huge magnitude of thrust change for a small change in servo angle. Maybe Mark can comment more on this...
@John Bond: I agree that it'd be awesome to see more DIY drones interest in variable pitch quads. You are absolutely right that quality components make a huge difference. In previous design iterations we used crappy components and ran into problems with vibration, control accuracy etc. Buying high-quality stuff is almost always worth it if you factor in the cost of your time. In terms of ESCs, we don't actually put them in governor mode. This way we can use them to control the quad in fixed-pitch mode on the fly. Changing the pitch has a huge effect on RPM though, so we've found that a small feed-forward term from pitch angle to ESC command works well. This method is of course open loop but seems to work fine in practice.
@Bradley Carr: We use a Vicon motion capture system to track the vehicle indoors. Low-level control is done onboard the quad and high-level control
Robert, right now I'm keeping the motor command and pitch relatively fixed. As the pitch is increased or decreased, the motor command is increased or decreased to compensate for the change in drag and keep the motor speed more or less constant. This gives the fastest thrust response by essentially eliminating the motor dynamics.
I've not considered ducted fans, but it sounds cool!
Mark, do you use motor speed control for stabilization or altitude control? Or are they relatively fixed?
Have you ever thought about using ducted fans with variable-area nozzles in a similar way?
Awesome! thanks Mark!
Sure. It is mechanically quite simple. I'm using Axi 2208/34 hollow shaft motors with MKS 480 servos. The variable-pitch actuation is accomplished using the EVPU (http://www.modelmotors.cz/index.php?page=101) from MsComposit. There are four Mikrocopter ESCs allowing us to update the motor speeds at 1kHz, while the servos accept commands at 333Hz. Everything is bolted to a sheet of carbon fiber/balsa sandwich board, with the UDB4 on top for control.
Mark
-
1
-
2
-
3
-
4
of 4 Next