Wheel grip stress test! (how good are my custom wheels?!?)
Vložit
- čas přidán 26. 08. 2024
- This week we stress test a few types of ant wheels to see how good they are
Previous test video:
• How much can a robot p...
Follow me to see what I'm up to:
Instagram - / _team_panic_
Regular schedule: Saturdays
Technically the force you can apply in a pulling contest is limited by the area pressure, so when you increase the contact patch you get more grip but less area pressure as the weight it’s distributed on a larger surface
Where larger wheels shine is if it’s a pushing contest where you wedge under the opponent, so the stronger you push the more weight you can get on the wheels up until you have the same area pressure ad the small wheels and so you gain more friction due to the better coefficient of a larger contact patch
This test should be testing friction coefficient of each contact pattern, the small dents on the pololu and the better rubber clearly has a better contact patch than the silicon-like slick rubber of the o-rings or the other
I think the best contact patches varies depending on arena floor, wood my benefit a lot more from groves than plastic, just like a dusty one may require studs, while a plastic ant or non spinner arena that’s supposedly clean of dust would be better with a porous rubber (rc car tyres, pololu, or rubber bands) to get the most grip
In the end however there is the constant that with N20 and 150g everything will just slide, unless you have a FWD grabber fully lifting the opponent, and at that point coefficient it’s not really important, with my 150g ant on rubberband wheels I can easily push 300-400g if weight if I wedge under it and lift it up a bit, totally shreds 1v1 force against pololu wheels
This was a wheelie good video. 😁 Wheel puns aside, maybe have a rough surface like sand paper (if possible). Or make the grooves flatter, thus more contact with the ground.
Love every single one of your videos, this channel is a huge treasure full of knowledge
That intro was cool! Did you get an editor? This video was really slick.
I’m looking forward to more of these interesting comparison videos.
Thanks!
Nope, just put more time and planning into the video layout. Hoping to keep this going for future videos
In the slow-mo, it looked to me like your silicon wheels had the most traction, which pulled so well that they made the back of the robot hit the ground and bounce
We all know flat wheels get most traction like slicks for a race car but we also know that a battle box is rarely clean and the grooves will help with debris and the slicks will just suck.
I love the scientific method approach to iterate your design. A couple of things I would be interested in seeing:
1. What was the maximum and minimum of the pulling force of those wheels? Your silicone wheels didn't do so well, but with that bouncing it was likely the average. I bet at moments your wheels had much more torque than others, just not consistently.
2. I would love to see a tug of war test. Both pulling force and pushing force. Having the weight of another robot that actually moves could change the results of the test.
3. Instead of treads that are "protrusions" from your wheel, what about a grooved wheel? That would eliminate done if the bouncing while still allowing for moments of high torque (at the sharp corner caused by the groove)
I don't know anything, just some ideas I had as an armchair quarterback 😉
Also, could you try a few different brands/types of silicone? Maybe a 2 part foam/flexible epoxy?
Tug of war testing is a popular suggestion, definitely needs to be done in a future video.
I've been meaning to try 2 part silicones, hopefully I'll get to them soon
would a hill climb possibly giving you a better idea of comparative performance even if its statistical value would be harder to calculate, the fact the you have more grip means that that you are transferring more energy to the ground but if you can move this could cause the wheels to store more elastic energy before losing grip launching the platform into the air were as the less grip causes the bots wheels to slip more allowing for a more constant force. Also if you increased the torque would you find ideal o ring amounts for different amounts of torque,
We're doing a decent because the robot has a tendency to wander off course, so the downwards slope helps carrel it in a straight line.
I like the thinking behind a climb, but in practice I think the bot would wander off course and crash into the sensor
Hey. Firstly, a larger contact area also means less force per square inch of contact. So smaller contact area equals more force, and also means that it can "dig in better", especially if the tyre material is soft.
It would make sense to measure "maximum static friction", which is the maximum possible friction before the tyre looses traction. And the kinetic friction.
Also, you said you used silicone for your own tyres? Cured silicone doesn't have lots of friction. Try actual rubber instead. You can get liquid latex online (it's used to make masks and FX). Try to print a tyre and just cover the contact area in a few layers of latex. Cured latex (rubber) is super sticky when clean! Real tyres are practically made of that stuff too.
I think many o-rings today don't actually use real rubber anymore (if they ever used that).
Thinking more about it. Theoretically, since maximum static friction is always greater than kinetic friction, it should make sense to limit the motor force to the maximum possible static friction, to prevent slipping. Once you slip, the friction goes way down.
I have been thinking about 2 part silicones and rubbers
Hadnt thought of liquid latex though, thats something to try for sure
Hey, loved the video, thanks for sharing. Do you have the numbers for static friction as well? Could be measured by ramping up throttle slowly and recording the peak force.
I dont, because the test chassis uses an arduino that pretends to be a RX and just outputs a full speed signal.
I could easily build a ramping function into it though, that would be a good test
@@TeamPanicRobotics I would highly recommend that. Lots of the "smarts" of nowaday's cars prevent slippage, since traction goes way down once you slip. Less force applied, actually equals more effect - like more pushing/pulling power. Also see my comment (which I made before I saw this comment) :)
Maybe try a herringbone gear type tread? With lots of small grooves that meet in the centre of the wheel.
Oooh herringbone tires is a cool idea
I compete in sumo robot category and wheel grip is essencial for us. Usually the wider wheels have the most power, but in the friction calculos area is not a parameter, only the normal force, which should mean that the thiner the wheel, the more pressure, but that's not the case. Whats your opinion about it? Also, I think this way os testing the grip os the tires is not very effective because clearly your wheel has more grip but not enough weight on them, what makes the robot jump, losing grip and pulling force (You're talking about it now lol, sorry i didn't wait long enough to make the comment). So maybe putting more weight in the robot helps.
Not sure how you could do that, but would some kind of "pips" (like on table tennis rackets) allow traction, while reducing bounces (and their loss) ?
Worth a shot for sure, might be difficult to mold thought
Your wide silicon wheels grip and bounce, so are breaking traction constantly. Look at how tractor pulling and sand rails work. They don't try and maintain traction as it bogs down (and would probably implode the transmission...) - there is an optimal level of wheel slip which creates the best forward momentum - ~10% from a quick Google...
Surface pressure may also be a thing. Wider wheels have less surface pressure over a given area so the rubber deforms less in contact with the floor and thus creates less grip.
Your silicon may have more ultimate grip, but the conditions mean it can't actually use it.
It's like axle tramp in a car when you spin the wheels making it judder and bounce
+1 on surface pressure, also probably worth checking the hardness of the silicone vs o-rings and the pololu (spelling?) tyres.
bit of a tangent, but something we did ages ago at work; we had some custom PU pads that were made with dual hardness. This might be an alternate way to make moulded-on tyres. These were two flat layers, and I think the bonding process required the first part to still be tacky while the second was poured, so maybe not the way easiest to make wheels, but there's possibilities there.
The harder part was shore D (like UHMW), the soft part was the squidgy shore A, can't recall the numbers offhand though. I think this could be more durable than the current over-moulded wheels as both parts are the same material, so they wouldn't try to delaminate when hit as much as having different materials.
They all have same surface pressure. Weight of the robot is more or less constant so wider wheels just spread that pressure over larger area. You'll have more contact points, but with less individual grip. This is exactly why 3 and 4 o-ring wheels have the same characteristics. The only thing that matters here are materials. From my experience silicone is not that sticky especially on smooth surfaces.
@@CockroachSlidy surface pressure is dictated by contact patch - it is a metric to allow comparison - in this case weight/area. Pressure cannot remain the same if one variable changes.
A wider wheel will have a bigger contact patch and for a given weight, a lower surface pressure. Think of how tank tracks spread the load over a wider area.
@@CockroachSlidy They have the same force (weight) acting on them, spreading the force out over a larger area gives lower pressure.
Spreading the same pressure over a larger area gives more force.
Pressure=Force/Area either way
Looks like I misunderstood the term. I meant that F = uN. So more area don't result in more grip if weight is same.
As a mere online mortal, the test is interesting but unrealistic. You rarely see a pulling robot and the string may be lifting the robot off the floor the harder it is pulled hence the bounce. using a push test may be more valid particularly with a slope robot wear the wheel will be pressed into the floor and the soft silicone could gain grip as it will deform more as more force is put through it. O rings are designed to squish a bit but not so much that the squeeze out of a joint, so you will reach a max contact area pretty quickly .
I wonder if attaching the robot to a draw slide would help. It'll lock the movement into a straight line without bouncing and skittering around the place.
And as for your silicone tyres. I would suggest removing the corners, so your cross section is shaped like a squashed U or semicircle with a middle flat
The problem with stopping the bounce is that it throws the test results out for a real world application.
Stopping the bounce would make a high grip tyre look like it would perform far better than it would in real life
Any ridged connection to the robot is going to add or subtract to the effective weight on the tires. Even the sting is going to have some impact when its bouncing. A better test would be pushing the scale like pushing another robot.
I would probably test on a surface with a controlled amount of sawdust on it to simulate the arena floor
Good point, dust and tiny plastic particles
Sorry this is likely a silly reponse, but isn't no pattern i.e. slicks, going to produced the most traction? There is not really much on your test-rig (or in a box) for a traction pattern to grip on to (and no water/debris to disperse..).
Possibly - tread allows the tyre surface to deform more so may be beneficial.
But I think in this case it may not improve his issues either way as ultimate grip perhaps isn't the problem - it's usable grip. It already bounces because it's gripping and releasing, so any more friction with the surface would exacerbate this...
@@TimInertiatic - to me it looks like there isn't enough surface contact due to the current design. At slow motion, it looks like large portions of the tyre are making little contact with the ground because of the tread. You are right that it appears to be gaining and losing traction in rapid succession, and that is causing the bounce. All the other tyres appear to be making much better contact. Frictional force between two objects does not depend on the area in contact.
I wonder if you wouldn't get less bumps and more traction if you were to slow down (significantly) the wheels. Perhaps you can try and trace with a servo tester at different settings ? Of course, you'd loose torque (less current), but torque is not the problem when the tire's splipping.
I'm using an arduino in the robot to send the RC signals, it would be super easy to slow down the wheels.
I'm adding it to the list to look into in the future
The intro is good. I would consider removing a couple of the VOs left in there.
Great production quality though!
My way of making wheels is 3D printing the hubs and finding a matching O ring.
Thanks!
Yeah I agree the VO can go, it sounded alright on its own but in a video its a bit weird
I'd say the next test needs to be a pratical one. You can crunch numbers all you want but the real proof is how it funtions. Build a pair as identical as possible, with adustable weight loads to compensate for the different wheel weights. Get them both responding to the same controller (or try dual weilding controllers lol), put them face to face and hold forward. The robot that loses ground has worse grip.
I feel something like this would give the most pratical data as to which is the best idea for wheels, and would set a bench mark for future designs of yours to get real world testing.
Just my thoughts. Might be interesting to see just how much extra grip/torque you get with a 4 wheel setup.
That's an interesting idea and might kill some of the weirdness of these tests.
I think the maximum pull force created just before they break traction would be a very interesting Stat to consider.
I'd need an actual force gauge to measure that, but I think that's where the series needs to go anyway
I hate to add more variables into your testing, but what about exploring differences in the durometer of silicone used?
Thats a good idea, but currently I am making these silicone wheels from a tube of bathroom sealant, as its cheaper and easier then 2 part silicone.
I want to start using 2 part silicone soon, so I'll keep this in mind
I recently bought some robot parts from the admin of my local Facebook group, and he told me he's been using RC car tyres. He claimed they're better than the silicons because they have flexible studs that help gain better traction.
Maybe you could make silicon tyre moulds with small studs to see if they make a difference.
BTW, good to see new vids from you again! 👍
RC tires do have a lot of grip, the problem is they are often the most expensive and heaviest option available.
Might be worth throwing them in the test rig though
I wonder if a tether like you have is actually throwing off your results, since the wheels may bounce/skid differently rotating away from something pulling them back, instead of toward something they push. To try to get it closer to actual fight conditions, maybe mount your digital scales vertically, and drive at them, or get a box and put progressively heavier weights in it to be pushed.
It's the same for all at least.
The weight also needs to be the same. The bot should always weigh 150g irrelevant of the wheel type, so lighter wheels need less weight.
If you are testing the grip levels, dragging the bot with non-rotating wheels down a slight incline would probably be a better test. It removes the bounce and variation...but it doesn't test the real world scenario, just grip in isolation, so not that useful
I tried a push in the first ever round of testing and the bouncing was still quite pronounced there as well, to the point that the value on the scale was jumping all over the place.
Might be worth retrying now I have the sensor set up
I think you should have done a 1 O-ring version also.
Thats a good idea, will do in the future
Ice racing wheel are really narrow to make more pressure on ground wider tire is for stability on cars mayby there this pressure on ground thing aswell
Im thinking maybe a tread pattern with thinner grooves so that the protruding bits are closer together, which then form a shape closer to a true circle.
Also if the cheap wheels are still getting the most traction, why not just print TPU hubs for those nice rubber treads (so that they are both durable and grippy)?
Great testing!
Are the diameter of the Pololu wheels larger? I think flexible nubs will give best grip, maybe 2mm diamond or round shapes, with not full sphere tips, but slightly rounded, so they absorb and hold some of the slip and deliver it more continuously while touching, even if they only flex maybe 1mm, it should help. Also offsetting them so they don't all grab at same time, like a diamond pattern in the tire, but tilted at an arbitrary angle.
Makes me want to print my own molds now!!! :)
Everything tested was 40mm diameter to try and keep some consistency
I like the flexible nubs idea, maybe I release my hub and mould blank for people to create their own
Instead of pulling perhaps you should try pushing. As for the pattern, I'm not sure how scale effects are relevant but I would suggest looking at types of car tires.
I did try pushing in the initial tests, but was pushing a scale and didnt get great readings
Its on my list to redo
@@TeamPanicRobotics Perhaps a spring can help? That way you might get a more averaged result
foam wheels are becoming more popular in 150g bots...
some people use latex to add grip to foam, some people use silicone, but i personally use this stuff called fantastic elastic... way more hard wearing but still very grippy.
maybe you can add foam wheels with different grip compounds into your next test. could also test tread patterns with the foam wheels.
in my testing, i found 2 foam wheels 34x10mm with fantastic elastic tyres and a 10x6mm nylon hub to weigh less than 5g.
A random, possibly impractical one. Certain chemicals make soft rubber handlebar grips (BMX ones as an example) go kind of sticky...
@@TimInertiatic from experience, sticky tyres work great for the first 10 seconds. then, after they have picked up the arena floor crap you have next to no grip.
@@TheDaddy187 very good point!
wow those are super light!
Might have to look into them as an option
Could you change the test setup to something more akin to a pitching machine - where you have 1 (or 2) wheels pushing a block into your weight sensor - this way you are checking how much force is imparted before they skid - ignoring the 'floppyness' of the robot side itself? after that you can possibly put them back on a robot to see what they do under actual combat environments - as it may be bench tests don't translate to robot tests due to differences of usage
Yeah bench tests not translating is why I started by building them into a robot,
Might be worth trying if only to see how much variance there is between "ideal" and combat situations
@@TeamPanicRobotics might also be worth it to see how reliable a bench test is vs combat - at which point having both could be useful as it might lead people to altering their combat setup to better use the bench test results
What about starting slow and measuring the force just before the wheels break free and start to bounce? Static friction should be higher to start with. I'm also wondering if some suspension in the robot might help. I'm actually considering building a piano wire chassis where the wheels can flex down when the robot is being picked up.
Pre friction brake would be great, unfortunately I am limited in how fast I can get data from the sensor.
Might have to look into a better sensor in the future
Hello. I'm new into building combat bots. If at all for sale I would like to acquire some fully built bots, kits, parts.. Wouldn't know if that would be of interest? Mainly looking to build 150g-1lb. Happy to pay for product and shipping.
Granted this is more a 3lb bot thing, but i use printed wheels, with pieces of cut up bicycle tire glued to them, so far i have not found anything that provides more grip
As for your traction pattern, go solid, a pair of slicks, maximum contact.
Screwed and glued bike tire is exceptional in larger weight classes, its what all my featherweights use
What shore rating do you use on the wheels you make yourself?
I dont know, its a tube of bathroom and glass silicone from a local hardware store, as its not intended for this use it doesnt have a listed shore rating
try boiling the o rings so there softer and grippier
Oh interesting I hadn't heard of this method, are there tutorials on it? Could be interesting to experiment with
@@TeamPanicRobotics fisherman will boil baits to make them softer, specifically frogs if your familiar with fishing, that way when a fish bites it will collapse the bait more allowing for a better chance to hook the fish. anyway just a random thought I had and I hope you try it out and have some fun with it. Thanks for responding your really a great youtuber for the antweight community and I love your content.
if nothing else, custom made wheels give you freedom to design them how you need them in the robots, using RC car wheels atleast gives you easy access to wheels
Oh for sure!
its soo handy to be able to make any size wheels to make a design work
I'd like to see a less scientific test and tie two of those bots together and pull each other to see who wins.
That's an interesting idea and might kill some of the weirdness of these tests.
@@TeamPanicRobotics honestly thinking about it you could even make a unibody 4 wheel drive setup and have them spin opposite and see which direction it drives to eliminate any variances the string brings.
I would have thought your grip is so chunky there is not much actually in touch with the ground.
Possible, I have designed smaller grip too, will have to try in the future