Thread Tools
Dec 03, 2019, 12:36 AM
There's magic in those wings !
khaled_abobakr's Avatar
Thread OP
Idea

Flight controller for Ornithopter !!


Hi Everyone,

I think the natural next big thing in the Ornithopter world should be using something like the popular flight controllers or gyros , of course totally different software and hardware to control the shape of each wing during the flapping cycle. Especially after seeing successfully flying servo driven Ornithopters. I think that could be a start, then advance to something more sophisticated like altering wingspan by changing the amount of hand wing sweep, of course the transition from gliding to flapping and vice versa will be 100% successful , turning by wing differential in flapping frequency, amount of twist, or wing sweep.... we'll see later which one will be most effective ... by trial and error ... it will be really good platform to use for research and development...

I can imagine a wing that has 3 servos ... one for flapping, one for sweeping, and one for twisting ...the flight controller should govern the harmony of these 3 servos during the flapping cycle for each wing... and control how they are positioned to steer the bird, climb or dive ....

I can imagine a less than 100 cm wingspan bird built so light like Kazu birds, but servo driven all around ... with that Orni-Harmonizer flight controller

Khaled
Sign up now
to remove ads between posts
Dec 03, 2019, 03:04 PM
Registered User
Dantiel's Avatar
i wrote code for one controller for accelerometer based on code for normal flying wing but didn't tested it yet... i think fine-tuning will be difficult.

especially while slow flapping it could help a lot to harmonize each stroke as well as stabilize bird against wind
Dec 03, 2019, 03:08 PM
Registered User
A flight controller isn't going to do you much good if there's nothing to control with it First we must develop the apparatus.

I'm still working on feathered wings, using 3-axis shoulder plus wing extension motion (elbow, wrist and fingers all move together), but it would be interesting to see other wing types on a 3-axis shoulder as well. Perhaps a rigid wing with Prandtl twist, using the shoulder twist motion to vary angle of attack throughout the flap cycle rather than relying on aeroelastic effects. May not be quite as energy efficient, but should give better control.

For the flap servo, I think cycloidal drive is the best reducer mechanism to use. Since it's relatively easy to fabricate (unless you want hardened steel discs), it can be highly integrated into the ornithopter structure, which greatly reduces its weight. For one thing the pin rings for both wings' cycloidals can be machined from a single piece of aluminum, eliminating the need for fasteners and a separate mounting plate to transfer torque between them. And instead of having a massive output shaft and coupling hardware, you can have the cycloidal discs turn the wing "scapula" directly, by integrating the cycloidal output roller pins into it. The geometry of the mechanism just happens to work very nicely for this.

A lot of people even 3D print cycloidal discs, but I think the highest ratio I've seen is 21:1, and I want 36:1. And my printer is not accurate enough even for low ratios, so I'm going to build a small CNC mill and do aluminum. I've been wanting to learn machining anyway.

ODrive is very good and reasonably affordable to control the flapping motors. But too big and expensive to use for all the motions on an RC ornithopter. I'll continue with my custom ESC firmware for controlling linear actuators. Fortunately they don't need holding torque, which is the main thing I wasn't sure how to go about implementing.

I'd really like to come up with a better wing twist actuator. I've still been working with this one in mind that I came up with last year https://www.rcgroups.com/forums/show...5#post40775833
But it's weak in the sweep-back direction, and I don't think there's any way around that without wasting a lot of energy in static tension.

I'd rather use one linear actuator for the wing sweep motion and something else for the humerus twist. It would help if I knew exactly how much torque it needs. There's not much space to work with, but it may be possible to wrangle a regular servo in there and have it turn the humerus tube with a cable wrapping system like what's used in this video: youtu.be/j1O0jkx97AU
I'll play around in Blender and see if I can come up with a good arrangement...
Last edited by dekutree64; Dec 03, 2019 at 03:40 PM.
Dec 04, 2019, 12:32 AM
There's magic in those wings !
khaled_abobakr's Avatar
Thread OP
Hi dekutree64

I was just tossing in idea around ... and I agree with you that a flight controller cannot be even made before the articulated wing is made first .... I'm happy that you're on this track .... I'll check the video and the thread links you posted ... and I hope one day smart people like you realize those dreams ... I can see that you've already started ... and that's good thing

Good luck
Khaled
Dec 04, 2019, 12:38 AM
There's magic in those wings !
khaled_abobakr's Avatar
Thread OP
I think a wing with multiple articulation points or axes for different movements need to start with only simple up and down flapping motion ... and then harmonize another wing motion with flapping ... one at a time ... until the whole wing is fully functional ... it will take a lot of time .. trial and error ... but's worth it ... I think it's difficult to get the harmony between those motion correct all together from the 1st time ...

Khaled
Dec 04, 2019, 01:52 AM
Registered User
Yeah, motion optimization will be a fun job. Maybe write a program with real-time adjustable settings, where each joint motion is operating on a sine wave at the flap frequency, and you can adjust amplitude, phase, and vertical offset for each one. And perhaps a setting to stretch one half of the wave and compress the other so you can do slower downstroke/faster upstroke. For a measure of flight efficiency, you could use average thrust force divided by average total power usage.

Either that or control the joints directly with sensor gloves and just play around until you find what works best. I'll also be doing haptic feedback of the torque on the flapping motors, and perhaps an accelerometer on the body to "feel" the thrust, so that will make it possible to hunt for optimum efficiency in an intuitive way even during free flight.

A wind tunnel with smoke streamlines would be ideal to study the airflow and get good thrust measurements, but for starters I plan to stick a post in the ground and mount a long rotating arm on it, with the bird on the end so I can fly in circles all day without danger of crashing That will also allow thrust measurement by monitoring the rotation rate of the arm. Probably not very accurate though, working outdoors in the wind, with friction on the arm's bearings, and the bird being forced to follow a circular path rather than a straight line. May need force measurements in the radial/vertical directions as well just so I can try to minimize them.
Dec 04, 2019, 07:10 AM
Registered User
Dantiel's Avatar
sweep and twist could possibly be controlled by one servo... like sweep back and twist up during upstroke, and then when wing is in up position sweep forward and twist down for downstroke. (but these may also be mixed depending on AoA, so 3 servos is better)

i think there have been some labs creating a fully functional and realistic bird movement robot, but i dun think it ever reached a state when it was light and strong enough to actually fly.

so there's two things, first a robot that is capable of flight and harmonize all movement, second to harmonize the harmonized motion with accelerometer/stabiliser that is getting difficult with many servos because all of them interact with CG of the bird... of course one could also only work on the rc input channels. anyway good birds/bird wings are already passively self stabilizing, but completing this with active movement might really hold firm for superior flight capabilities.
Dec 05, 2019, 04:16 AM
There's magic in those wings !
khaled_abobakr's Avatar
Thread OP
Quote:
Originally Posted by dekutree64
Yeah, motion optimization will be a fun job. Maybe write a program with real-time adjustable settings, where each joint motion is operating on a sine wave at the flap frequency, and you can adjust amplitude, phase, and vertical offset for each one. And perhaps a setting to stretch one half of the wave and compress the other so you can do slower downstroke/faster upstroke. For a measure of flight efficiency, you could use average thrust force divided by average total power usage.

Either that or control the joints directly with sensor gloves and just play around until you find what works best. I'll also be doing haptic feedback of the torque on the flapping motors, and perhaps an accelerometer on the body to "feel" the thrust, so that will make it possible to hunt for optimum efficiency in an intuitive way even during free flight.

A wind tunnel with smoke streamlines would be ideal to study the airflow and get good thrust measurements, but for starters I plan to stick a post in the ground and mount a long rotating arm on it, with the bird on the end so I can fly in circles all day without danger of crashing That will also allow thrust measurement by monitoring the rotation rate of the arm. Probably not very accurate though, working outdoors in the wind, with friction on the arm's bearings, and the bird being forced to follow a circular path rather than a straight line. May need force measurements in the radial/vertical directions as well just so I can try to minimize them.
Dekutree64 .... That's the spirit ... Hahaha .... You just took the 1st step towards realizing this project .... I really enjoyed your suggestions ..

Khaled
Dec 05, 2019, 04:23 AM
There's magic in those wings !
khaled_abobakr's Avatar
Thread OP
Quote:
Originally Posted by Dantiel
sweep and twist could possibly be controlled by one servo... like sweep back and twist up during upstroke, and then when wing is in up position sweep forward and twist down for downstroke. (but these may also be mixed depending on AoA, so 3 servos is better)

i think there have been some labs creating a fully functional and realistic bird movement robot, but i dun think it ever reached a state when it was light and strong enough to actually fly.

so there's two things, first a robot that is capable of flight and harmonize all movement, second to harmonize the harmonized motion with accelerometer/stabiliser that is getting difficult with many servos because all of them interact with CG of the bird... of course one could also only work on the rc input channels. anyway good birds/bird wings are already passively self stabilizing, but completing this with active movement might really hold firm for superior flight capabilities.
I like that Dantiel ... this is why a flight computer/ controller is a must ... just like multi-rotors ... it's impossible to fly without one ... and in our case ... the software will be far sophisticated because of the variables during each flapping cycle ...as I said it has to be really simple at the beginning then activate / harmonize one joint at a time ... and before that like you said a wing must be created so it's durable enough to carry those servos and additional articulation mechanism yet be light enough to fly ... and that will be one hell of a challenge ....

Khaled
Dec 10, 2019, 03:22 AM
Registered User
Ok, I've finally got a 3-axis shoulder design that I'm satisfied with. I may need to adjust it for a larger twist servo, but I think the basic architecture is good. This is designed for the common SG92R servo, which has 2.5kg.cm torque, although that will be at least doubled by the cable wrapping system to turn the humerus tube (mentioned in my previous post, but not shown in the 3D model).

The main shell structure is what I call the scapula. You can think of it as a highly modified output shaft for the cycloidal gearbox The top end sticks into the cycloidal (through that large bearing), and the bottom end is supported by another bearing, but this one goes inside the scapula and over a short smaller diameter hollow shaft on the ornithopter frame. This bottom shaft (not shown in the model) will be removable to allow for assembling the whole thing. The wires for the actuators will pass through it.

Inside the scapula is a linear actuator for controlling the wing sweep motion. It will be made from a Hextronik 10g outrunner with its shaft replaced by a partially threaded 2mm steel rod, some brass tubes, and some 3D printed parts. The ESC for it (running my custom firmware for servo-style control) will need to be stuffed in there somewhere, but I think there's enough space. If necessary the actuator can be scooted toward one side to make more room.

The humerus uses a tube-in-tube structure, with the outer 16mm carbon tube being part of the wing, and the inner 8mm tube being part of the shoulder. The outer tube rotates over the inner tube on bearings.

The sweep axis shaft is a 17mm carbon tube with a hole drilled through it for the 8mm tube to pass through. Around the 17mm tube is a 3D printed part which provides the connection point for the actuator piston, mounting area for the twist servo, and a flat surface for the end of the 16mm humerus tube to bear against (with a steel washer as a basic thrust bearing).

The range of wing sweep motion is about 70-75 degrees, with the forward extreme being straight out to the side. That should be good for biomimetic bird wings, but to adapt it for a more traditional wing spar where straight sideways is the neutral position, you'll need to modify the scapula to move the sweep axis toward the humerus opening direction, to give better reach around the cycloidal reducer. That will put more thrust force on the shoulder bearings when flapping, but shouldn't be too bad.


Quick Reply
Message:

Thread Tools

Similar Threads
Category Thread Thread Starter Forum Replies Last Post
Discussion Best simulator for Ornithopter simulation and control? Ajith Kumar Ornithopters 3 Aug 16, 2019 01:58 AM
Help! Control and Wiring for Servo Based Ornithopter quadconnect Ornithopters 2 Dec 29, 2018 12:49 AM
Discussion Hacking a Flight controller for ornithopter nataraja18 Ornithopters 12 Aug 01, 2017 08:29 AM
Discussion Simple Control Surface question for Ornithopters CosmicDog Ornithopters 7 Oct 08, 2016 12:42 AM