Vibrating Circuit Robot
Vložit
- čas přidán 9. 08. 2020
- In this video I show my journey in making FlexBot, an attempt to make a simplified robot made from a flexible circuit!
Learn more about Altium 365: www.altium.com/altium-365
PCB Design Software that I use: carlbugeja.com/altium
Free Version of Altium's Pro Version: www.altium.com/circuitmaker
Thank you PCBWay for manufacturing my flexible PCBs: www.pcbway.com
Support my work on Patreon: / carlbugeja
Open Source Gerber Files:
www.hackster.io/carlbugeja/fl...
hackaday.io/project/174265-fl...
First PaperBot Concept: • Paper Robot Experiment
My Other PCB Robot Concept: • PCB Robot MKIII - Fina...
Cameras/Tools I use (affiliate links):
Main Camera: amzn.to/30tRgY7
Second Camera: amzn.to/2ZLm2wy
Light Setup: amzn.to/3jtqckA
Magnifying Lamp: amzn.to/3jrcaQv
Vernier Caliper: amzn.to/3eTju3I
Multi-meter: amzn.to/3hq6B2R
Digital Milligram Scale: amzn.to/30HIw1X
My Social Media:
/ carlbugeja
/ bugejacarl
/ bugeja.carl
/ bugeja.carl
/ carl-bugeja-0b922a135
Music:
Cutting It Close - Audio Library
Fringe Trimmed Boots - Audio Library
Malt Shop Bop - Malt Shop Bop by Kevin MacLeod is licensed under a Creative Commons Attribution license (creativecommons.org/licenses/...)
Source: incompetech.com/music/royalty-...
Artist: incompetech.com/ - Věda a technologie
Big fan all the way from South Africa.
Put the “leg flaps” on isolated legs to allow you more precise lateral control. Add a variable power source to adjust your upper limits of your pwm. This way you can control how far the legs must extend, still with great current control. reducing the dependence on the hinge tension. And yeah, you figured the turning problem out. That was simply your center of mass. Very nice work!
Lol when when he said he's been working on it for a long time he wasn't kidding
I wonder if we put a series of PCBs in a straight line, would it make the robot better at moving straight. It would be like a caterpillar PCB.
Having the section below the coils that rides against the table be two points instead of a a line would probably make it less sensitive to mass distribution and to having the bend line be perfectly parallel with the contact line. You could also make the back contact a shorter line segment. This would essentially make the robot stand on a tripod instead of two lines.
I was thinking just the same thing!
Agree! I was thinking about make both sides stand on like a zigzag instead on a line.. But I am not sure how many points
This could be tested with the MK-II, just by cutting away a few bits of the straight edge. It does not take much, i would guess even just cutting away 0.2mm with an exacto knife would show you some results. I'm not sure, but i guess having the contact points to the floor surface as far away from each other as possible would probably be the best.
Now this is what I call innovation
Since you asked, two suggestions:
1.Make the hinge a square or "box" hinge, with two folds. This will reduce the "spreading" caused by the PCB wanting to go flat.
2.Put a drops of hot glue, rubber cement, or just something with a greater surface area and higher friction on the "foot" so it's vibrations can be more easily transferred to motion.
wow never been so early intresting concept looking forwards to a project with lasers and flex pcbs would be epic, love your videos
Instead of folding the PCB inwards, make 2 PCB and join them at the upper part, that way the unpowered position would be together, and the powered position would be open. Also, because the magnets have a thickness, they would keep the 2 PCBs at an angle making the robot stable.
that's what i tried using the glue for :) the solder pad idea should work as well
@@CarlBugeja Kinda, when you glued them there was still a distance between the magnets and the PCB. What I mean is for them to lie compleatly against each other. And only when powered they would separate. This way they should be more powerful.
Anyway, thanks for your repply :)
Seems to me it might be the fact that the hinge is one piece (i.e it has only one big leg, it has to bend the single piece to steer, this would mean the two coils would just fight each other when turning) if you split the two coils the robot would have two independent legs and more yaw authority, you could add a tiny gyro and have the software run a feedback loop that would keep it going straight even with imbalances in the system. Awesome project.
This is a good idea imo
Oh man, I love your work and it's been quite nice to see the FlexBot walk. I think you should do the changes you mentioned at the end of the video and conclude this project with a big success :) Keep up the great work.
Awesome!
love your projects bro!
every PCB is interesting
you're an inspiration and a rich source of knowledge, please keep going!!
Some ideas you could try:
- Introduce a phase offset between the two coils to try turning.
- Make the two halves a "B" and "D" shape (better yet, a "W" and "V" shape). If the edge has three sharp points, you have known and consistent points of contact. With the flat sides, you're basically making contact at random points based on the roughness of the table and the variations in the PCB's edge.
- Make the angle as tight as possible, so it doesn't have to fight its own weight.
- Move the glue/solder away from the bend in the PCB, so you don't have to make as tight of a bend radius or fight so much springback.
yes, I like this! I will probably never make something like this, but I enjoy the recaps of your journey.
You can crease the plastic of the flexible PCB like a cardboard box. I used to be a tool maker for print finishers where we made tools to crease plastic as i imagine would work with your PCB. Instead of one single fold/hinge, have two. Make the distance between these crease folds the thickness of your magnets. This way you can make the PCB stay in a folded shape and have the coils parallel to the magnets.
Have you considered some kind of closed loop control? With a gyroscope, this way you could control the pulse rate to get straight motion without the problems of manufacture variation
That's what I thought, an accelerometer to detect the walk angle deviation, and compensate for it in software.
what about a compass?
@@altrombone1775 A compass is more like a low pass absolute reference. A gyroscope has more high frequeny dynamic response. This is why IMU generally fuses compass and gyroscope data.
I like your way of facing these challenges. You are always thinking of new ways and solutions. When I think about whether you should continue or stop here, my advice would be to continue. Not only because I am excited to see how this robot will turn out, but because I think you will come up with some new great ideas in the process. Really love your work!
thankyou carl .your videos are really inspiring
Great job bro- my favorite tech CZcamsr 🙌🏼
Nice engineering and smart work all the best with your project
Nice experiment!
these are always interesting to do just to see if you can do it. Great to see that you made progress near the end by putting a more balanced magnet and bigger battery onto it!!
Make a caterpillar with these! There's room for a soft robot here.
Loved it bro
To solve the issue of "posture", calculate the distance you should have between the two parts of the pcb sides at rest (when not driven and standing up right), in order to have it stand up with those two pcb parts vertical. that distance needs to be the distance occupied by the to permanent magnets plus the expected distance to the coils. at the hinge side you need to either solder, or glue some solid piece of plastic or similar, that can act as the rigid distance provider. I believe once you did this, you will just need to add at the bottom of the pcbs some "shoes" for the robot, in order to allow better adherence (probably some 2x2mm pieces of rubber or similarly adherent material). Looking forward to seeing the next iteration of your pcb robot. :)
your current design is essentially a 4-legged design (the wide back has two 'corners', and the front has two 'corners') so your robot will rarely ever sit perfectly flat, which is another part of what's potentially causing it to not move reliably straight even on glass.
Adding a small nub in the back for the circuitry/battery side to rest on a centered 'foot', and splitting the coiled area in the front into two separate legs, would probably result in better movement, just like how a 3-legged table is always more stable than a 4-legged table :)
This robot flexes your skill
It seems like the crease is an uncontrolled variable. Instead of tweaking the material thickness, you could eliminate the crease entirely. With a long fold area and some tabs and slots, the fold could have a large bend radius (making a small loop instead of a crease), and the the two sides could be kept in proximity by the tabs and slots. This way there is a solid mechanical feature maintaining the distance between the coil and the magnet. You could reverse the mechanism so the panels are normally closed instead of normally open, which may allow for more design options here as well.
Lovely research. Flexi bot can be use to make robot butterfly. Keep rolling.
I loved the butterfly video of yours
Superb i'm doing it. It's what to use my time for
Intresting concept
What you want is two different currents to drive the coils. The speed of the movement will then be able to change depending on moving forward or backwards. If you move a side forward fast then back slow, it will move more than if both the same speed. The duty cycle will have minimal impact since you are moving both directions with same speed so sliding friction is the same.
Ur genious buddy🔥 awesome this is one of the great projects I have seen in my life can you make project on rfid with custom rfid coil
My noob hints (for what they worth) : It could help if the coil side parts that touch the ground have some rubber feets. One little foot for each coil, so 2 feet.
As your robot, sideview, is *basicaly* a triangle, with only one freedom degree (radial rotation on the folded axe), I think :
- pushing the coil away from the magnet will lift the feet up of the ground (rotation around the fold) before the robot hit the ground again (inertia, then gravity)
- pulling the coil to the magnet will grab the rubberized feet on the ground, acting as a hook, so the folding robot will move forward the rear part which has no rubber feet, sliding it forward.
The weight distribution between coil side (grabing side) and magnet side (sliding one) will be essential for this to work.
i.ibb.co/R2jpS1T/Flexible-Robot.png
always make a great video,
The problem is the flexible PCB material, it will always naturally want to return to its original flat state. The easiest solution would be to cut out as much of the PCB material in the hinge area as possible and replace it with something else that will push the two halves of the PCB together, instead of apart. (This is also the reason why turning left or right isn't working, because you can only utilize the actuation force of the coils in one direction and the difference between left and right isn't big enough unless you can utilize the reverse actuation of the other coil.) The hardest part is you have to find a spring material that will match the force of the coils as closely as possible and that has a low dampening ratio. I would recommend natural rubber (like a thin rubber band). For the design I would recommend making the hinge a bit longer and making a few holes along the hinge where you can tread the rubber band trough. Something like this basically:
____ ____
/ \___ __/ \
| o|_|o |
\ o o /
/ o _ o \
| __o| |o__ |
\____/ \____/
And then use either glue or tape to bond the rubber bands to the PCB.
You can try to add small tabs to act as foot
Your videos are just what I need. :)
I have an idea, add two dragging flexible arms so you can steer it's direction.
_Hi Carl, this problem is quite easy to solve._
_Put 2 bends close together so that the halves of the circuit are closer to parallel._ 👍
I was thinking the same thing
Maybe make a square hinge the same thickness as the magnets?
interesting idea 🤔
@@CarlBugeja have you tried putting copper in the bend area of the flex pcb? When you bend the copper it should help the edge keep its angle, rather than trying to open up again.
I think another option could be to attach magnets to the side with the coils as well. No PCB redesign needed and the glue in the hinge would be obsolete.
I'm not sure about the permeability of magnets as they are already magnetized/saturated, but I'd guess it's significantly better than air and can thus result in larger forces at the same coil current.
The coils would gain moving mass/ store more kinetic energy. This will drastically reduce the resonant frequency and drastically increase the elastic/kinetic energy stored at a given oscillation amplitude, which may be beneficial
I was thinking the same too, I think this is the best solution for your problem
Great 👍👍
I admire your level of engineering. That being said, you could try use rubber band instead of electrical tape to hold legs together, maybe tiny spring can work to. Also, TARS has 4 sections, can walk on 2, 3 or rotate like a wheel on 4
I think you should keep going, it's a nice project
On the note of the baseball robot, a tube launched robot and/or drone would be intresting. Use a pnuematic "potato gun" to launch either the ball, or a larger device.
(also just a return of that series would be fun)
A throwable/ "hit-able" baseball bot could be cool too!
Wow
It's awesome
I still love the flexible PCB robot concept and think it'll work! Would love to see a Mark III !
Your video is so cool
No ideas or recommendations, just wanna say your videos are great
Приятно смотреть идея )
I’d definitely up go for the solder pads, and make 2 or 3 per side so you can get the most symmetry and repeatability of the connection. I’d also consider having two flexible parts with a magnet in the middle, such that both the front and rear legs push off the same magnet, which just hangs in the middle. Adding some rigid pointed/serrated parts that interface with the ground to the bottom of the pads might be a good way to ensure that you’re getting contact with the ground only where the “feet” are. Might also increase friction too much though, so you’d need to do some testing.
Because this will likely require quite a lot of trial and error, I’d advise just connecting a dummy weight on the robots and having the motor driver and comms off-board, connected with some thin wire.
Great !!!
what an interesting concept lol
If you create a longer hinge area, you can have 2 folds that would allow wider angles on each bend.
Keep it up brother! Your videos really are an inspiration to me. Wish I could give some actual technical advice but I'm only halfway done my Electrical Engineer degree :p
It's so cool.
This may not be related to the folding (and alignment) issues, but I was thinking about switching from the permanent magnets solution to a full 4 inductors solution, so that you can control each pair individually. This implies a bit more complex circuitry of course (full h bridge for the already existing coils as you did, and a simple low side mosfet to control the other coil always with the same polarity), but you should theoretically get rid of the alignment issues with this, and the overall design might be lighter. Also, this will give you more fine control over the magnetic field I suppose. However, coil pairs might need to become really close to one another to work as expected... Having a double bend (as suggested in the comments) will certainly help as well!
this is such a cool robot.
i think if u make the coil and circuit pcbs separate and then just use a stiff U-shapred wire to act as both an electrical connection as well as a hinge to connect both sides together, u would be able to control the angle of bend.
the stiffner added in between the pcbs is also making the pcb more flexible nd is making it go back into its original shape, i think u should remove it so that u can deform the plastic more easily in the bent shape.
That's starting to look pretty cool! I was about to suggest splitting the front between the coils, between the two coils, but I see Tainton Hutton already suggested that. That would allow more movement of each leg without so much impact on the other leg. Also, could you put a couple of holes near the bottom of each leg and run a piece of thread between the bottom of the leg and the bottom (the part that touches the table) of the electronics board? Then you could adjust how far the legs moved just by pulling more thread through and tying a knot to keep it from going through the hole. Have you considered some kind of rubber or foam along the edges of the board where they touch the table, to give it a little more traction? Or a tiny roller at the bottom of each leg, that only rolls one way? Maybe make them out of the spring pins they use to hold the band on a watch? They're spring-loaded at the ends, so you could compress them to snap them through a couple of tiny loops of wire to act as axles. This is starting to get really interesting. Keep it up! :)
I think your PCB have quite huge part dragging against the surface.
I mean,
·.____.· ·.____.· this is the lower part of the actuators.
It can be reduced in size from the bottom and that might help in reducing the force robot have to move against.
Agree! I think it would be much better if it was like a zigzag
@@CarlBugeja Yess
Yeah, i think trying a shallower angle might be worthwhile. I really like your explorations :)
Maybe not zigzag but only two points contacting.
I think adding a second hinge might solve your problem. So, instead of have a V shape (looking from the side), you're going to have a U shape.
This seems like a really cool project that's got your head around the software. I totally dislike the cloud based software when you've got crap internet though.
Hey Carl. Try 2 batteries, one on each coil. the independent mass on each coil should give you better control as you will have individual momentum on each leg.
Ok hear me out; only have the coils themselves on flexible pcb, have the magnets and other components on solid pcb.
Excellent video and series!👍
i would have used some fishing line (or other light thread) to stop the flex part to open up too far. Make some holes near the bottom of the PCBs, then put a string between the two sides that holds it together, kinda like you tryed with the electrical tape. If the holes are small enough, a little knot in the thread on the outside of the PCB would be enough to hold it place and limit the movement, and it would be easy to adjust the maximum opening angle.
You could make the joint area of two flex pcbs as slotted structure or you can use the same concept of flex cable that runs from one pcb to the other and may be make a dummy flex cable as a joint fpr two pcbs. This will increase flexibility in the joint area... Cheers!
It may need a pair of feet to adjust the friction. The foot doesn't have to be fancy, just a plastic tab would do, change the curve of it would change the pace.
If you make the hinge M shaped, you might be able to glue to outer folds fixed and control how closed it wants to be with the middle fold, I think
I love your robots. Coul it be possible to charge battery with the coils?
Успехов!
Also could try heating the PCB while it's folded, and letting it cool down in that folded position.
Yeah, use a Heat gun!
What you could do is to model a slightly thicker bit of PCB, then using a knife, to cut a slit in the upper most layer, reducing the angle
Did we travel back in time? 🤔
No its a repost, maybe 👻😜
Yea as per previous comments, a square U shaped hinge with flexible wire and solder pads or holes should hold or reduce the angle. Also cutting or seperating the middle part should give you better control and an option a sort of trimming which is similarly found in RC toys & build. Like that the current flow could be reduced in one and maintain direction instead of turning around. 🤷🏻♂️
I am sure you understand differential drive robot, some improvement you can make to you current design is to split the coil(leg) into seperated coil(legs) instead of joining them together in your current design hence you can better steer the robot by actuating with different frequency.
1. try to make two legs separably 2. try to make copper lines along the fold
Cool project so far :-)
I'm not shure if it is better, when the pcb is not flat where touching the ground, but some sort of individual legs, as you mentioned. Perhaps some tests with a rigid wood prototype can show the direction of the project. Perhaps it is better to have only 3 or 4 touching points to the ground then two regions. This might also help with uneven surfaces.
Just a thought... because your forward plane is a straight flat line, i can see how it would randomly favor one direction or the other. How about trying to slightly curve the face plane so its an arc, or even crease it for a triangular front plane... but i think just an arc will give it more natural direction balance, and not the randomness of a flat plane.
So, do you think about hinge between coils and PCB like a door hinge? I mean that add some holes in it for a better flexibility.
You could try to use two different coils faced together, to try to get rid of the (heavy) magnet
nice
I like your idea and i think i can help with your problem i don't think the problem in the electrical design or the size of the magnet it's the weight distribution and you have to make some kind of legs like glue two little rubber at the bottom of the coil side and you need weight on this side too to give extra grip on the rubber and for the hinge you can cut it and glue it just from the top while you stash them face to face you can use the soldering iron for it hope this will help
You should sell this in a long slim 'bullet' formfactor 😏
Rubber bands. Maybe try them to limit the movement, so instead of just stiffness you get elasticity? Should also be more efficient maybe? Just attaching them might be hard. Also the last attempt seems quite promising, but maybe add "feet"? It seems like the random direction changes are because the robot gets stuck on the surface.
maybe you could put holes at the edges of the PCB, and tie the two parts tobether with thread so they can't slide apart. then the thread lengths could also be changed to maybe adjust it til it moves straight. or just one thread in the centre to keep them together.
The battery should be mounted low and at the rear for stability and rigidity. The front needs a slot to create to front legs
To help it go straight better you may want to work on balancing the mass of the circuits to that there isn't too mush on either side. That may be contributing to your problem with going straight.
Good
Pretty cool Idea and work man!
As everyone suggested, separate legs and close-loop control (using magnetometer gyro etc) would really go a long to improve things.
Also how about soldering springs on the hinge?
Why not cut the coil section vertically from center so that both coils can move independently? Also, for more control it's better not to rely on plastic or paper. Try to use repulsion/attraction for bi directional movement. It's an outstanding efforts from you. I wish you success.
I know this is more about your PCB actuators, but could a robot like this also work with (stacked) piezo actuators?
I think having a little bigger gap between that two side can make better movements
Maybe have the folding hinges on the sides, and remove the long folding hinge across the top?
when you folded it, did you try heating the edge and letting it cool to remold it, as when they are produced they are flat ;)
Put a small spring from a clicky ball pen or something in there to keep the two flaps closer?
I bet you could make a different type of propulsion using the equivalent of insect wings formed from thin PCB. Many pairs along the length of an inverted-U board, some of the board folded down for legs, and some straight out as wings. Make the hinge point at the leading edge so the trailing part of the wing twists slightly during the flap.
Great video, you could use an IMU to make him walk straight, I don't know if the magnet or coil affect him.
If you have enough power to move forward, just try to move one side forward while the other backward. You can't do it at the exact same time, but with a little time delay. That should result in rotating CC and also CCW.
That's what I tested when tried to turn left and right 🙂
Have a look at inchworm robots. They are quite similar. There is the RRL laboratory at EPFL who work on something similar (but with a different type of actuation principle).
what i think is you need to make on side longer than the other and maybe make a bind in the longer end . good luck
You may try to separate the two coils to give independent movement
I feel this could be modified to be used for swarm robotics research.