Taming the Tail-Sitter: Hover to Forward Flight Explained
ฝัง
- เผยแพร่เมื่อ 2 มิ.ย. 2024
- Thanks to PCBWay for sponsoring this video--check out their website for an instant quote on custom PCBs, 3D printing, or even CNC machined parts: www.pcbway.com/?from=Nicholas
In theory, tailsitter VTOL aircraft are incredibly simple--just add a wing to something like a quadcopter, pitch it over 90 degrees, and cash in on the sweet, sweet efficiency benefits of forward flight. But there are actually 4 key challenges to address in the flight controller for a smooth transition between hover and forward flight.
Patreon: / nicholasrehm
dRehmFlight VTOL flight controller code used in this video: github.com/nickrehm/dRehmFlight
Tailsitter specific code: www.rcgroups.com/forums/showp...
Drone flight controllers that support transitioning VTOL configurations often simplify the complex task of performing a transition to the flick of a switch. This is awesome for people who like stuff that just works, but terrible for people that like to understand how things work. So how do you actually tell a flight controller to perform a VTOL transition? Rather, how does a flight controller tell the airplane to perform a transition? Tailsitter VTOLs need extra care in the flight controller through their transition, mainly because the definitions of roll and yaw control with respect to the body flip between hover and forward flight. This requires careful remapping of stabilized controls out to the motors and servos, in addition to a flip of the fundamental aircraft frame of reference. Another key to a smooth transition is tuning PID gains separately for each flight mode, and ensuring a seamless and smooth fade between all of these flight mode differences. This video covers how I did this using a little bit of custom code in my opensource flight controller, dRehmFlight. If you found this video interesting, don’t forget to subscribe so you can be up to date on future projects like this!
Intro: 00:00
Flight Controller Basics: 00:54
Problem 1 - Switching Roll & Yaw: 02:58
Problem 2 - Switching Frame of Reference: 04:30
Problem 3 - Gain Staging: 06:58
Problem 4 - Smoothing it Out: 08:04
Conclusions & Crashes: 09:45
#VTOL #Tailsitter - วิทยาศาสตร์และเทคโนโลยี
I want to demystify these sorts of drone concepts all the way down to the actual implementation. Did seeing the code help, or bore you?
It was super interesting to see the code! Makes it all more clear and interesting :)
Seeing the code was helpful, thanks!
Love seeing the code..... it takes a while for me to understand it.... never boring... I have spent hrs and hrs on your projects.
This was a well thought out application challenges/educational video; however, I cannot help to think if you had simply added two rudders behind the motor thrust-line so regardless of the flight mode you had same control inputs. Any thoughts?
Very interesting. I need to learn to code though.
This is probably the most cogent explanation I've seen about tailsitter flight dynamics. Breaking down the problems one step at a time, plus not being afraid to show code, makes this super easy to understand. Excited to see your next build!
Hearing this gives me the awesome feeling of "mission accomplished", thanks a ton!
@@NicholasRehm 🫡 Least I can do when my latest bicopter is powered by dRehmFlight!
Seeing each step was fantastic! I'm especially impressed he was willing to fly all of the intermediate steps that he knew were going to fail. It's so easy to just think through these problems and skip to what you assume is the finished solution, but actually testing the physical response of using a binary transition or not changing the forward flight gains really drives the lesson home.
"Hey look, it's RC test flight"😀. Great explanation of the transition problem.
this video actually took work to make and offers something unique. all RC test flight does anymore is drive around in a boat.
@@dronefootage2778inaccurate
@@dronefootage2778tbh tho im still here for the boat stuff (i started watching in like 2017 for the solar plane project 🤩)
@@dronefootage2778I’m wondering where your great output of high effort work is?
@@GeahkBurchill i just watch youtube
Awesome video. Just a word of advice or suggestion. When displaying the stability properties of a quad rotor or 2 rotor vehicle sitting in a stable hover, try pulling the bottom end toward you rather than pushing the bottom away from you.
This way, as the vehicle rotates from your physical input, the thrust from the rotors will send the vehicle away from you instead of tilting the props at your face.
Ask me how I learned that and I'll show you the scar and point at OG Phantom 3 that did it. (Bastard DJI)
Yeah, for real... safety rule when demoing stability or GPS lock, always pull the quad skids or lowest part toward you, never push a quad away... they just come back at you harder :D LOL
Very good safety advice... I admit I was taking some risks in order to get the shot on camera haha. You'd think I'd have learned by now given my run in with one of those props in my last video
😬
The thought didn't even occur to me, thanks for putting this out there
I’ve been wanting to make a tailsitter for a while now! Great video!!! Keep it up!
Thanks! Hope this inspires you to follow through with it
@@NicholasRehmwhat parameters on ardupilot should we pay attention to in order to smooth and tune our quad tailsitters? Please let me know!
This is the most impressive "unimpressive" piece of foam I've watched in a long while.
The fact that you give such a good explanation into how and why you did things, makes this video to be into a whole other level. It passes from being an entertainment video to be a super good educational video + still being fun to watch. I'm exited to see the next project.
Thanks for the kind words
So you basically need 2 FC profiles, different PID values, reference orientation & controls. Neat!
Also, the fade part is smart!
so smart that you put a timeline on yr advertisement.. I watched the whole ad rather than clicking ahead, because I knew the length.
Yep, I think it's time to breakout the drehmflight board again and play with it some more! You really have made something special here with your software and videos. I appreciate you taking the time to share this stuff with the world
Thank you for adding the code in the video. It helps me to understand. I can't wait to see the next video.
Nice! It's always cool to see people who are able to do physical design/builds, complemented with coding skills to solve both problem domains. Really looking forward to your large, folding design!
Thanks! Hoping to have that one flying any day now
@@NicholasRehmI hope it goes well!
That tease of a reconfiguring quad-to-fixed-wing looks VERY cool.
Glad you was able to hear what I was screaming through the screen from the beginning of your video about you need a factor changing the weight of control and two different controllers for diffenent modes. :D
Very nice video ! Especially great that you let us in on the engineering process. ❤ this !
The only thing your “unimpressive piece of foam” needs to be my dream aircraft is a camera, vtx, and code that pivots the camera mount 90 degrees during flight mode transition. Your next project looks absolutely stunning! I can’t wait to see it! Thanks again for all your hard work and for sharing it with us. 🙏
That sounds like it would be a sweet setup. Check out my buddy peter's mini-qbit for something eerily similar th-cam.com/video/W0tthPSNnRE/w-d-xo.html
I suppose a second camera and a PWM-controlled switch would be about the same weight as a servo and linkage, with less mechanical complexity. If the camera switch signal was sent as part of the flight mode transition function, it would make the transition even simpler. Thanks for the video suggestion. 👍
Amazing in-depth vid, thanks for inspiring others!!!
Well done... Flies and transitions much smother now.
Nicholas, your videos are great and this one is particularly excellent. That was a great demonstration of engineering synthesis and how to systematically go about solving multiple problems. It was great that you were able to show how the VTOL behaved under the various steps of the design process.
Thanks for following along!
I don't recall how I stumbled upon this channel but I'm happy I did. The way you explained all that gibberish was impressive. Keep doing what you're doing!
You make such brilliant, groundbreaking videos with some nice engineering work too. We’ll done Nicholas, another one in the can.👍
Thanks so much!!
fabulous video. i really like the progression: stepping through each problem on the path one at a time, and placing the viewer in the mode of expecting each change to be the last one, but then exposing the remaining issues to address
Hey thanks! It was a fun process to work through and I'm happy I could share
You are brilliant. Everyone wants to see you build a single-seater VTOL machine capable of 1 or 2 hours of flight time, will essential avionics (com, nav, weather), and some safety features (enclosed blades or jet ; crash parachute...) ; plus a pivoting nacelle - that would be THE airborne motorcycle
That would be sweet
Great technical overview! I purchased Horizon Hobbies Xvert 5 years ago. Great VTOL and performs like your plane for transitions. This plane is a ton of fun to fly and very inexpensive. Keep it up! Love your videos.
Lots of fun tech packed into that plane!
Congratulations. Very impressive.
This is so GOOD, Nick! 👍👍
I get how a smooth transition between hover and forward flight is better, but dang that sharp transition looks kinda badass, too.
Perfect..... are you putting the raw code up like you did with the F35? This would be awesome to have... or is it documented somewhere. cheers
I think I'll be doing a write up on the rcgroups thread and I'll post it there--excited to see what you cook up! Hope you're doing well
That would be awesome....... I have tried a few tail sitters but nothing worked hence the excitement and comment.. Still flying the F35, made a few tweaks to the transition timing so its dialed in and does not dip down when going into FFF..... I have the Nemo set up ready just waiting to assemble the printed parts.. also have another teensy ready to be set up for a small F18...... I was wondering how big a plane you could use this board on as I have a few larger foamies that I want to try it on just for stabilisation.... as for me I just retired 3 months ago and we have been travelling and just relaxing..... if you want to see some of the plane stuff I have been doing since retiring feel free to check out my YTch ... I hope all is well there with you, stay safe and keep up the great work.. BTW I still check in the RCG thread to keep up with what is going on... cheers from downunder
@@NicholasRehm
Very well explained! Outstanding! I'd like to fly a foam VTOL like this!
Great project! Thanks for the video, Ill be waiting for more!
Hey thanks!
You have a great talent to explains things
Appreciate it!
Awesome work! Incredible what a single - very intelligent and hard-working - individual can do. Many thanks for the fun and inspiration.
Thanks for the kind words!
No problema señor@@NicholasRehm. Keep up the good work!
Awesome video, excellent well done.
Bruh..... this science of yours seems a lot like magic 🤯 (nice work!)
Hopefully a little less mysterious once its broken down!
You have some AWESOME videos! Thank you!
Hey thanks!!
As a programmer, as soon as you had the violent transition working I was like, ok nice, the final step is just an easy lerp from hover to forward!
great explanation, detailed and clear.
dude, i always love your videos. Now they are getting funny too with your crazy humor😍🤩
Hey, I know you aren't impressed with your "piece of foam", I am impressed. I've been trying to get a plane working but struggling with the balance because I have older, heavier equipment.
Basically, another awesome job! Well done.
Thanks for the comment, keep at it with your project!
Love your videos. You're a great engineer.
Thanks!!
Amazing video, thank you very much!!! Can't wait for the next one...
Flight dynamics and controls is such a challenging problem. So many important details!
Great clear walk-thru of the basics leading up to achieving stable transitions. Being able to make reasonably smooth transitions to level flight without airspeed or angle of attack sensors is pretty amazing.
The ultimate test would be to fly the tail sitter through a loop, or barrel roll ... involving transitions to hover at various points in the maneuver.
That would be fun. Wish I had more time to dig a little deeper
Looking forward to your next Tailsitter larger project.
Really interesting video!
Thank you man! Great to do :)
A very informative video, many thanks.
Not sure if you realized, but at the end with the scaling factor you reinvented “fuzzy logic control”
I guess you're kinda right!
Great explanation
Reminds me of something :)
I was planning to try the gradual transition sometime, but using a little simple sin/cos scaling. Seems to me that the relation to gravity magnitude is what matters, and that does not change linearly with pitch angle.
Maybe not gravity... It's true the gravity stays constant but I think the distinction is because regardless of pitch angle, air provides the same control authority from a control surface regardless of the direction that the plane is flying. Air gives the same resistance regardless of whether you fly up or sideways.
great work
Love how to showed the code and talked about how. Peeps leave out the details so thanks for the info cause I want to do this shit but I ain’t that so any info is awesome!!
I made this video for curious people exactly like you 😄
It's an interesting coincidence that the TH-cam silver play button and quadcopters both have 4 corners🤔👀
Great video, keep it up!
i am impressed man , pls tell us more about this nice piece of foam. :D
Thanks for the arduino version added to the RCGroups page... now to sort the wiring out..... if you only knew the amount of times I tried to set up a tailsitter years ago using other flight controllers....this seems too simple to be true.... now to dust off a few old wings.... and print some more glue on motor mounts..... cheers
Wow that works really well
This is an awesome project! I love things that are mechanically simple and reliable yet work really well.
One thing I don't agree with is the "quadcopter" control layout when in hover mode. As an rc pilot with 30 years experience. My brain naturally would tell me to fly this plane shaped vehicle with an airplane control layout. Using the yaw stick to control yaw as if it was pointing straight up, and roll to twist around the vertical axis.
This would also make the configuration much simpler.
Agreed, sometimes my brain confuses me too being primarily an airplane guy. Thanks for the comment!
The code is very clear and simple and well written. Thanks for making that available, it helps see how all this works. Have you done any more on the spinning 3-wing drone. I love the way it looks in straight flight. You left a teaser at the end of "part3" but I don't see anything since and that was a year ago.
great video.
This is an awesome project and walkthru for the "hover to forward" flight problem, I followed your thoughts the whole way and it is so satisfying! One thing I'm not sure about is at 8:20, did you set the fade transition time on a fixed timer, or is it something else that's controlling/adjusting how fast or slow the fade phase should be?
Thanks for answering my question and for creating such awesome content, keep up the amazing work!
It's on a fixed "timer"--in reality it's just rate-limiting how much that variable can increment with each flight controller loop if that makes sense. I tune that increment size which directly corresponds to total time from 0 --> 1
@@NicholasRehm as your flight controller loop has a fixed frequency, a fixed rate also results in a fixed transition time... or where am i wrong?
Well that freaked me out.
I was across the room listening to the end of the video and I hear (seeming to come from someone in the room) “hey look it’s RCTF” and of course the next thing I hear is Daniel’s voice as a video I had started yesterday auto-played. I was utterly confused for about five seconds. Well done lol
That's hilarious hahah
This tailSitter looks so much fun!
I want to build one ❤
I found that you provide the final code over at Patron!
You should totally rebundle this video as a build video, and market it as: a cheap sub 250 project you can build with your kids and have a great intro to programming 🎉
This is basically kiwico without the parts.
And you buy working code by signing up for Patron 😊
"not ai unless you are an Investor" good one
this is indeed the easiest way to transition. Cant wait for the big project!
Do a short video on the hardware and programming environment you used for your development, Nice Video
Vtol amazing. Great movie.
This would be a great way to get beginners flying.
"unless you're an investor" 😂
I have built my own "tail sitter" I used iNAV in heading hold mode. I find the transition of roll to yaw and vice versa confusing, although given in my case, I don't need it to land in hover; given I have an advantage that my craft can do high alpha up to a hover.
IMO the advantage to roll and yaw switching is in tilt rotors. Still a feat to do it all on your own code.
Very interesting!
So cool!
Thanks!!
From Ireland thanks
Great project, results and walkthrough of your process, and learnings. What software do you use for the visuals/diagrams? Cheers!
PowerPoint haha
This settles it, I'm making a Drehmflight FC PCB thats actually made properly. Last time I tried it really wasn't up to spec, but seeing what you can do with it tells me it's worth it to go again.
Check out @michaelrechtin's 3d printed quad project...he made a pretty sweet pcb breakout and I have a few on my desk now. Here is another someone made: github.com/joerenteria/dRehmFlight-PCB
cool tech 👍
You could also call this one the aileron sitter. If I was doing this I may want a bit of stand-off so it was not actually landing on the control surfaces.
Loved the video. I just finished a project using the ESP32-S2. It was a big one. 1.1MB in Flash.
I know even the ESP32-S3 doesn't have the power of a Teensy 4 (@600MHz), but does it have the power to run your code decently?
Would it be the clock speed (240MHz) that's not fast enough? Was ESP32 available when you wrote your code? Or you just went with the best and biggest horsepower?
6:23 looks yummy! Soon I hope...
Some people have ported the code over to the esp32-s2 and said it runs just fine. The teensy 4 has STUPID compute overhead for this relatively simple code, I just like having the option if I want to load it up with other stuff :)
@@NicholasRehm Great. I hate to learn yet another hardware.
@@NicholasRehm interesting, also did they use Arduino or ESP-IDF? Also i am not that surprised it runs well on ESP32-s2 since unlike an arduino uno or RP2040 it has a floating point math unit and for the type of math that you are useing it is a great gain in speed. Also remember NOT to use core0 when useing the WIFI/BT since it may lag important code running on it .
The tensey 4 was a huge overkill to begin with (unless you wanted to ad video/photo to SD)
I remember designs for carrier based tailsitters for the US navy in old popular mechanics magazines. Without fly-by-wire they would have been death traps.
So neat. Is this something someone could pickup and flash to a tailsitter or are there bonus steps?
Very cool.
I wonder what happens if you use pitch input as a transition fader, for 90 deg pitch use all hover, for 0 deg pitch use all airplane control settings, and mix proportionally for all pitches in between. Maybe you won't even need a dedicated switch, but instead, push the stick forward to transition.
I mean, remove the notion of two different modes altogether and use a single control law for both, blending two different control laws based on the current pitch?
Unfortunately the ambiguities/singularities of euler angles complicates this
@@NicholasRehm Interesting, thanks.
Hi, I'm new, trying out a tail sitter VTOL. Am I required to use a CW+CCW motor combination? Or I can just use a CW+CW to make this work?
I would really like to know. I'm so inspired to make one of these. Thanks in advance!
Great video. Love how you explained different aspects of the problem the approach to resolve it. I am thinking of buying a teensy and build one of these. Will it work on a nano goblin which has only one motor or will need a lot of tuning and changes to the code?
Sure you can throw it on a nano goblin to stabilize forward flight, but I don't think you'll be able to hover it
@@NicholasRehm thanks for the reply. I was more interested in the hover actually to make the nano goblin a vtol. Any specific reason why this would not work?
@@NicholasRehm will putting this on a heewing t1 ranger hover it? That has 2 motors but a real elevator instead of elevons. I am assuming no. But just confirming.
Ooh new video
Very interesting. VTOLs are on my list for ages now... I did some contraptions years ago, but the electronics/hardware were a limiting factor, not now tho. Next thing, deciding on type of VTOL, less dead weight + less moving parts = grater efficiency... so this type looks very promising, but as usual, it got downsides...
I'm very interested where are you going with this, keep us posted :-)
Primary downside of this particular design is susceptibility to wind, further complicated by reliance on motor thrust for control authority (ask me about how easy it is to control while descending at low throttle….). I’ve got a design I’m cooking up that I think solves some of these issues at the expense of a little more mechanical complexity!
@@NicholasRehm"little more mechanical complexity" - hhehhh, that's where problems and fun begins, good luck m8!
wow it is insane
Would it help to read forward airspeed using a pitol tube, set a target airspeed, and base your transitional mixing on those.
If anyone would like to try designing and programming this in a game I recommend Main Assembly
Your channel are so good that I unsubscribed and subscribed again! Nice project, keep them coming! Thank you!
Thanks!😁
"The sensor fusion algorithm is not AI, unless you're an investor in which case it's definitely AI." 👏😂
Where has this been all my life? I'm putting this control scheme in my AIM-7 for next years FF!
YOU'RE THE SIDEWINDER GUY????
@NicholasRehm I had the big white missile, with the dual drone motors, it was technically an aim 7 sparrow
Would it not be easier to auto-transition the controls depending on pitch angle instead of a switch input?
yawservo = yaw*sin(pitch) + ail*cos(pitch) type of thing?
Obviously it would cripple aerobatic flight, but that is not the goal i think.
I am impressed by the work done. Is firmware based on Ardupilot/BF/iNav or something else? Can I try this FW on my aircraft?
It is my own flight controller code running on an arduino-compatible microcontroller. Check out the link the description to the GitHub repo
i liked the violent mode switch more it made a cool sound
Fantastic video, is there a version of the code specifically for this tail sitter, I would like to try building one of my own?
I'll probably post the modified version from this project in my rcgroups thread sometime soon
Incredible work! Do you have any thought as to the increase in range in airplane mode. Thanks.
definitely more
@nicholasrehm thinking about making your own flight controller or a variant of the teensy board with the IMU integrated now that you are working with PCBWay?
amazing
Hi. I want to ask a question about your previous video: "Is THIS the Most Versatile Aircraft Design? (Forward Flight Testing) - Part 2". Could you please explain how you solved camera rotation problem? I want to do it in my project but I did not understand how. Thanks for your help.
A bearing and air brake
is "yaw" spinning about the vertical axis wrt to the earth? or wrt the plane. new to all this but i was under the impression it was spinning around an axis perpendicular the wings surface
"trust me, I'm an engineer."😂