Join my DISCORD! discord.gg/fc6MedG7eW INSTAGRAM: instagram.com/xrobotsuk TWITTER: twitter.com/xrobotsuk MASTODON: fosstodon.org/@XRobotsUK

@justpaulo

Жыл бұрын

8:00 I don't understand your math. I believe you are wrong. I think the horizontal track must rotate 2x faster than the V shaped ones (and not 1.5x). PS: I keep trying to paste a link to a picture showing my geometry argument, but the comment is automatically deleted.

@jamesbruton

Жыл бұрын

@@justpaulo Well it works and that's the way I've built all omni robots

@justpaulo

Жыл бұрын

@@jamesbruton I know, I see all your videos... You do an impressive work by the way! On those other projects I also commented something, or tryed to, but b/c of links the comment gets deleted. I am going to try to share with you a picture of my google drive that I think it shows my point. Basically lets imagine that @ 7:42 the robot is moving right with speed 2. The right track points in a 60º angle and the little wheels in that track point in the -30º direction. For that track to move smoothly 2 to the right, the little wheels need to move √3 in the -30º direction and the track to move 1 in the 60º direction (it's the famous 1, 2 , √3 triangle). Therefore the speed of that track is 1/2 of the horizontal upper one.

@jamesbruton

Жыл бұрын

@@justpaulo No it isn't. The horizontal track runs faster otherwise it spins as it drives.

@justpaulo

Жыл бұрын

@@jamesbruton But that is what I'm saying.... I'm saying the horizontal track must go 2x faster though (not 1.5x).

An idea for more intuitive omni-robot control: Put a compass in your controller, so it knows which direction you're facing (which it can assume is also facing the robot). Put another compass in the robot, so it knows which way *it's* facing as well. Say you're facing north, and push the stick to the right - the robot will figure out it has to travel east, and work out what power it needs to put to which treads to go that way. It'd be more intuitive because all you're doing is telling it to go left, right, away from or towards you; you could also add automatic rotation, so it can always wind up facing whichever direction gives it the most efficient/powerful travel (perhaps with manual rotation so you can still wiggle it left and right to get over extrusions). It's an omni-directional robot - you shouldn't have to care which arbitrary direction is "forward", that's its job! :)

@nerys71

Жыл бұрын

Oh yes some of the lower end drones have a similar feature called heading lock it doesn't assume you're facing the drone but it always assumes you're facing the same direction which was surprisingly effective once you figured it out

@edmills9736

Жыл бұрын

​@@nerys71 "Headless" mode on drones.

@jackmino729

Жыл бұрын

You could probably combine this with some other automation so it can follow you around

@TalynStormcrow

Жыл бұрын

Great minds think alike, I just suggested similar without seeing your comment first. 😊

@Beef4Dinner22

Жыл бұрын

For anyone wanting to try something like this, the sensor you would be looking for is a magnetometer (specifically you would probably want a 3 axis magnetometer), which measures magnetic fields and can tell you which way is magnetic North.

Do you have a system that manages all your prints and printers? Seems like a lot to keep ...track... of.

@its_nihar

Жыл бұрын

Yes I have had the same doubt. Yeah basically if you do em all manually i think they'll probably take a month or you don't sleep 🤣.

@Dthwsh7

Жыл бұрын

THATS A GOOD ONE

@Spiker985Studios

Жыл бұрын

I'd presume Octoprint would be able to do so, but alas I haven't used it

@Esablaka

Жыл бұрын

Not sure about lulzbot, but several of the large 3d printer manufacturers offer "Printfarm software/support". There you basically have all your printers piped into a program and whenever there is an object to be printed the printer who currently is free will start automatically (and which has the configuration like filament etc.) . There are even manufacturers with hardware options that clear the bed automatically and store the parts in a bin so you can print continuously without needing to interact with it at all. Creality, prusa, flashforge etc. all offer this functionality as far as I know. One can also set it up oneself via octoprint or a software like Simplyprint .

@HarveyShankar

Жыл бұрын

I've heard he keeps them in a tank.

Amazing job taking all those suggestions into the new design. I guess what we learned was, a 3 tread tank has more issues than benefits

@ulysses4536

Жыл бұрын

I would argue that we needed 2 less iterations out of total 2 to prove that it's not a good tank design.

@anonymouspersonthefake

Жыл бұрын

@@ulysses4536 it still might make a nice remote control toy

@ronnetgrazer362

Жыл бұрын

@@ulysses4536 Waste of plastic, time and network capacity. I'm done with this channel.

Bro, your robotic inventions will make the world change

Your videos are a wonderful example of how 3D printing can be used as a prototyping tool

The DoD should hire you! I never could imagine someone could make the tank LESS EFFICIENT!

Really enjoy how you casually critique your work. Similar to skepticism in a good scientist. It’s cool to see your reworks of projects, that process alone is fascinating to see. Keep up the awesome videos! 👍

Holy, just started getting recommended your content again. Can't believe it's been 7 years! Most consistently great content!

Making the swivel axis motorized could mean you can lift up that back track when it gets stuck which could help with clearance

@widgget

Жыл бұрын

i was thinking this while watching. would make for some interesting looking movements like being able to stand up on the tips of the tracks.

@wurstelei1356

Жыл бұрын

The drive mode where 2 tracks move half speed of the 3rd track is already very good at climbing things but your idea still is very smart. The angle of the tracks on ground would change with your idea thou. It also would work if you rotate the tracks 45 degrees around their long side so the tracks form ramps to the sides. You have to take obstacles driving backwards in this case.

I think those custom treads with just 2 tracks total would work incredible. One of the main benifits of tank tracks is that it can turn on the spot (ie like omni wheels). The third track seems incredibly silly after this video haha. Now a 4 track version (2 front, 2 back) would be what would be awesome for James to scale up to full size

@Mysda_

Жыл бұрын

Then it wouldn't have any sideway control with just 2 parallel tracks.

@stalkerfromvoronezh4493

Жыл бұрын

@@Mysda_ But 3 tracks fuck up all idea of tank tracks.

@Mysda_

Жыл бұрын

@@stalkerfromvoronezh4493 I agree, feel like 2 tracks with mecanum 45° wheels could be a better option. But I don't know if that can turn or not

@2009dudeman

Жыл бұрын

@@Mysda_ That wouldn't work. mecanum wheels require groups. Two tracks would behave very strangely. It could drive forwards or back and pivot in one spot, but any attempt at differential drive would result in unwanted side shift in addition to the turn you intended. We did try a 3 wheel mecanum setup and it kind of worked, but it was not very good and had some strange behavioral quirks that made it essentially unusable. You can have more than 4 independent wheels, it just makes the math more complicated and you can no longer just use the single drive equations for each motor (sin functions) you now need to keep track of two coordinate systems and calculate the vectors for each wheel individually to determine the required wheel speed.

@Mysda_

Жыл бұрын

@@2009dudeman Any place to see that? I'm interested! I was really wondering because screw-propelled tanks seams similar enough to the concept, and they actually are able to go forward/backward, rotate and also strafe with only 2 motors. Also yes, it can get pretty complicated. I was brainstorming about a neural network capable to adapt to any omniwheel setup on the fly a while ago, never ended up making it.

Minor nitpick: The groves on the wheels go the wrong way. Those little wheels work the opposite way than normal wheels, where you want traction when driving them. Here you need trac(k)tion when they are not rolling. So the profile needs to be turned 90°. As this is probably not great for 3D printing adding spiral grooves in both directions will also work.

when i was in service, i was a cavalry scout on a bradley. this whole project has my brain screaming "HERESY!" i'm glad you came to the same conclusions i did that this would not scale up well at all. not to mention the problems with using such a configuration in the real world. see, the way tracked vehicles brake is track tension. with the wheels on the track, on unlevel ground, this design would not be able to stop. it would also have large control problems going up or down all but the most shallow of inclines.

A+ for design and build F for practicality

If you took the rubber wheels within the tracks (at the bottom) and made them much thicker and with holes in them, like airless tires, you would also achieve a suspension effect, which tanks have. In fact, its so important for tanks that they have their own unique type of suspension. Also what youre trying to build vaguely resembles the omnidirectional treadmill for VR

I really like seeing the iterations over time.

Usually there's only one teethed wheel per track, the others are just smooth road wheels. There are alignment teeth on the track that stick out into the grooves of road wheels to keep it in the center. The tracks are really just to help spread out the weight of the vehicle. Omnidirectional Super Swamper!

To me it looks like you were able to pass the original obstacle because you significantly scaled up the size of your tracks and kept the size of your original obstacle the same as it was before. I would have liked to see it climb the obstacle with it being scaled up the same amount. The loose metal bars in the end didn't show much, they were just sliding around there and you used lower ones to get higher up. I'm almost certain that if you take your original tracks and obstacle and add a similar ramp to it, then it would probably climb up as well.

googley eyes always helps me keep track of where the things are facing

Amazing job, I love your design.

Prototyping with printers is great to see proof of concept. But don't forget scaling is important. The reason full-size tanks drive over things is in big part to the weight. But, as the weight gets bigger, things like small wheel parts could be too easily snapped off by the torque physics. Those police bomb robots climb stairs well, partly due to weight. If they were made to be super light, they might just slide down the stairs, flip over, or not even get started. I love the simple design of a three-wheel per axel, hand-truck dolly for climbing and driving.

I didn't expect the track is struggling to move sideways, awesome build!!!!

To fix your issue regarding the obstacles, I would add a fourth track to make it a diamond tank instead of a triangle tank. Then hypothetically the two rear tracks should have just as easy a time getting over the obstacle as the front tracks.

You can make the wheels on the edges of the tracks larger without them interfering with each other by putting them on every other link rather than every link. The limit then will be the frames holding the sprocket and idler axles. The track frames could use spring loading to prevent them flipping over. This is such a bizarre cool project. I'm glad it came up in my feed. NASA and ESA need to hire you for planetary rover designing.

Super. I like anything track related.

Perhaps if the tank tracks were a curved profile, instead of a flat profile, it might be better at climbing over obstacles with the third track. However I think that this would cause the track to have a very large bend radius or not bend at all. If we could create a trapedoizal shape of 3 pieces/track instead in this axis as well, and give it small torsion springs to keep it in that position normally, while keeping the drive gear at the bends wider than the middle piece, which would cause the piece of track to flatten out and hence still keep a small radius. Yes, it introduces more complexities but I would really love to see that in a v3 someday🙂🙂

@Thekilleroftanks

Жыл бұрын

that would defeat the point of tracks. the main reason for tracks is to help spread the weight, its climbing ability is mainly due to a tanks suspension (case in point theres a video of an m3 Lee having trouble going up 13 inches while a panther kinda just glides up and past the problem) and nothing to do with the actual tracks.

@harkrits.nagpal6505

Жыл бұрын

@@Thekilleroftanks I mean that is already the case in this video, since the wheels are big they are the ones contacting the ground.

he is the mister maker of 3d printing

Just put some low-friction ramps leading out from the centre, down to the underside of each track. Then the inside edges of the tracks won't get caught going up steps.

nice and clear transition. Good job bro.

I had a robot for vex in 8th grade that had omni-directional capability, and what I did to get over the issue of remembering where the front was was to convert the joystick input from X and Y to a vector, add the direction the robot was facing using a calibrated IMU, then converted that vector into information for each of the driving wheels. This meant the when I pushed the stick up, the robot always drove away, no matter which direction it was facing.

You are an excellent engineer… now all you need is excellent driving skills 😂 💜

This a good example of how omni wheels only work well on a flat surface. Even if you make them into tank treads, they are even worse on uneven surfaces.

Woah, you're a smart and talented engineer!

What a redesign!

This would make an impressive "howl's moving castle"

If you want to maintain that sort of a profile what you need to be able to do is to tilt the tracks but not the way you're tilting them You're tilting them on their rotational access coming directly out of the body instead you need to tilt them on pitch so when you are driving using two tracks you need to pitch the dead track so that it's angled up in the direction you're traveling and when you reverse you need to pivot it the other way You should be able to software link that I would guess but I'm not a software guy You don't really need to pivot on the radio access too much because you can't really conform to the shape of what you're running over anyway since your track units are too stiff You don't have anywhere near enough moving in the tracks along there track rolling access So you need to do is stop the dead track from getting stuck just have it pivot up so when you're moving forward that dead track would pivot so that you're basically riding on one edge of the rolling wheels on the edge of the track and then the portion of the track closest to the body would be angled up like 30° this way when it came into contact with something it would just push right over it

Quite an impressive project. While I have absolutely no idea how to build a robot, I'm leaving a like for keeping it open source. Good job mate!

Much larger idler wheels would probably help with the step issue

You should make the front be facing the long side of one track so that the back track doesnt get stuck onto whatever its climbing

Imagine a real life military purpose triangle tank. What would its purposes be? This is just a thrown out guess but would a triangle tank be more efficient in an urban war environment. Because a triangle tank could turn extremely quickly enabling it to have better movement / agility property's. Also a triangle tank could still move about if one tred was to be blown off. Just putting it out there.

@jort93z

Жыл бұрын

inb4 you get stuck on every curb cause of the tiny wheels. Don't think the tiny wheels are something that can be solved while still keeping the tank track shape.

@LordDragox412

Жыл бұрын

@@jort93z Just replace the tiny wheels with tiny tanks that have wheels replaced with tiny tanks that have wheels replaced with tiny tanks that have ( ... x infinity), problem solved!

@laz272727

Жыл бұрын

Tracked vehicles can already spin on a dime and smaller urban robots tend to have four tracks anyway, so this scheme is still worse than the default.

@xaytana

Жыл бұрын

@@laz272727 This and these omni-track designs lose the benefit of effective surface area, the entire point of using a tracked vehicle. An ideal omni-track, which would essentially be a perpendicular track around the main linkages, would be overly complex and still have sacrifices. And grousers are impractical on omni-tracks due to the fact that you'd either need a segmented blade that can be rotated per perpendicular track, or you'd have to settle with using a studded design that sacrifices grouser effectiveness for simplicity; especially considering grousers will help with climbing capabilities, plus various capabilities in varying terrain types. Omni-tanks are impractical no matter how you look at them, and for additional maneuverability that really only replaces a one-point turn and a three-point repositioning, with some additional vector efficiency that's more than likely negligible at the end of the day; plus screw tanks exist for higher maneuverability, though their only downside is that they're less effective on hard surfaces, at least not without a segmented and positionable screw. There's just no point to omni-tanks, really the only practical use for an omni-track would be for something like an omni-directional walkable surface where you don't have to necessarily worry about a sprocket drive which simplifies linkage design greatly, and you don't have to worry about other factors other than keeping the surface fairly flat; which circles back to the optimal design being a perpendicular track around the main track's linkages. I really don't see this series evolving into proper design, and I really don't see it evolving into displaying the one practical use-case of an omni-track. Yet, as usual, 90% of the comment section doesn't seem to understand the basics of mechanisms nor have the logical thought to figure things out, and we all know James will never point these things out himself.

@nuclearmedicineman6270

Жыл бұрын

An agile tank still can't dodge an RPG fired down the street.

Very cool. Outdoor Test would be nice.

Hey James, an addition I would recommend would be to create an indicator of what direction is forward, the simplest I can think of is just a 3D printed part that just attaches to the front of the drone, like an arrow with high contrast, or it could be as fancy as you decide, I am no engineer, an I certainly know nothing about electronics, but if you were to want a fancy mechanism that shows your bearing, I should not elaborate, as I tried, but ended up scrapping that thought, knowing I would be horribly wrong as my inexperience shows

Here is an idea for how to stop the back tree from getting stuck while traversing over the plank. Try putting a slope on the inside of the tread unit (not the moving treads) so that the tank will just slide over it.

I feel like a way to deal with the Radius Issue might be to, instead of having the treads flat, have them flare upwards at the ends. That way you can make the radii of the small wheels bigger without having to touch the radius of the sprockets

@jort93z

Жыл бұрын

i don't think thats really helpful. sure, the radius of the sprocket wouldn't change, but the radius of the track would. That's like simply making the track thicker.

@MegaGaming11

Жыл бұрын

@@jort93z Fair enough. There's also the problem that you'd have to make the track pieces longer to accomodate for the flare as well, so in a vacuum it sounds like a good idea, but actually implementing that solution would be a design nightmare. Part of me still wants to go with the "raise the small wheels so that they interact with the floor differently" route, but, as you stated, that *does* still increase the radius of the track. Doing that would also change how the tracks interact with the surface(s) they drive on, so it's not a solution that's feasible.

@jort93z

Жыл бұрын

@@MegaGaming11 I mean, if you curve it up, essentially it would only drive on the small wheels, driving only on the rims. That sort of removes part of the benefits of the tracks in the first place. The benefit of tracks is that the load is spread over a large surface area.

@MegaGaming11

Жыл бұрын

@@jort93z True! a potential solution to that particular problem would to just be to extrude the pieces downwards to match with the now curved/extended part of the tracks, but that presents its own issues and also reintroduces the original issue of "Ya can't increase the radius of the small wheels without increasing the radius of the track" if it even got rid of it in the first place.

@jort93z

Жыл бұрын

@@MegaGaming11 "would to just be to extrude the pieces downwards to match with the now curved/extended part of the tracks" ...And at that point you are back to basically just making thicker tracks. Which would let you increase the size of the small wheels.

Beautiful

Great project! It seems like the whole driving experience could be improved by implementing driver-centric steering by adding a gyro and adjusting where forward is based on the gyro value.

An alternative to making the little wheels bigger is to either angle the tracks toward the outside edge lifting the inside off the ground or making the tank level itself so that it lifts the back track up to the level of the higher surface

An alternative to the hobby axels are 1.6mm TIG wire.

Amazing project! I love the idea of making a tank that can move omnidirectionally! I may look at adapting the design for my own project. I personally believe that an H track with a centrally located drop down omniwheel would work better for getting over obstacles. The triangle design seems to have trouble getting more than one tread over an obstacle due to the angle of attack being 60° off of the direction of travel. Plus an H track would be considerably easier to program. Just put the strafing controls on the x axis, forwards/backwards movement on the y axis, then rotation on the rotational axis of the joystick. Overall, love the idea!

Interesting experience, thanks. James keep up the good work. :) :) :)

Great idea 👏👏👏

Could you put a skid/guide 90° to the track between the core and the track to lead the back track up to the obstacle?

Neat proof of concept. I would try adding a spar at the front and back of each tread that goes into the main body and put something like a ball joint at each connection point. It would inhibit the rotation of the treads enough so that they couldn't flip.

I think with some coding, you could impart a slow spin to the system while still moving in the direction requested. This would self regulate the rear track snag problem. However, you wouldn't know which direction is forward since 'forward' would be constantly rotating with the spin. You could mount a turret on top that always points 'forward'. And, if you go as far as a turret, you might add a camera so that you get a robot POV. The flipping track issue might be solved with some pins and rubber bands so that as track starts to flip, the rubber bands start producing more and more counter force to bring it back to neutral/level.

You totaly need to combine this bot with that head-bot you mentioned !

Hey, i used you model in a robotics competition at my school, thanks i got pretty high.

man said good at getting over things they are huge gun on tracks

This would make such a fun droid design. I really love the way this moves. Will we see any continuation of the B2-EMO build? I really loved that droid, and your designs for him were great.

You still make tanks better than the British in WW2 despite all its flaws

Seems like this is one of those concepts that's just doomed to never work out. The perpendicular wheels are more or less fundamentally incompatible with purpose of tracks.

For version 3 perhaps you could try a push pull cable suspension for the 3 tracks. The cable could redistribute force when one side leaves the ground. This is what we used on my senior project building a lunar rover. I’ve got some videos on my channel

2 more improvements: Make it so the tracks can't swivel too far and make it so the tracks can swivel up as one piece that way you don't have to make the wheels bigger.

If you could power the pivot point too, to rotate the track then you could raise the back of the unit to stop the 3rd track catching and also lift the whole body to stop it grounding.

30 degree angle on the edge of each track plate. Three wheels on each track plate wheels closer together almost touching. Should help the next wheel bump up. And fit a servo to force flatten each track module to horizon slowly . So it can Force the tank to climb and get a different foor print

Rotate constantly and overlay the absolute direction of movement 😉

#Sweet! Looks like you just need to flip the driving direction to keep the non moving wheel in front and push it over the obstacle

I think the big issue you cant really overcome is that trapezoidal tracks only help when the track is going head on. When collides side on, the edge of the track being flat on the ground can't really go over anything. Still cool.

@TalynStormcrow

Жыл бұрын

This can be fixed by keeping the tank always slightly rotating.

It's crazy how it only took you 12 minutes to build this!

Seems like there needs to be a suspension system of some form so the front 2 tracks can pull the back one over. The issue seems to be from the rigidity of the mounting than any driving issue. Overall cool build.

Thoughts for getting that last back track up Any kind of slanted piece you can put down from the main body towards the track to act like a ramp.

KZread recommendations came through, for once! lol This was really neat!

To solve the direction issue you could use a gyroscope and use that to make a north-east-south-west and drive based on that instead of the directions of the tank. Basically using the orientation to the room/starting position to decide which way to go. Problem is that you would have to stay on the same side/place in the room or relative to the starting position.

Pivoting the tracks is a bad idea. That basically turns the third track into a plow that gets stuck on everything.

Try doing it with a bulldozer track design which would allow two larger diameter sprockets at the ends. That would greatly reduce friction. If you want a slight tilt up at each end look up a T34 tank.

Perhaps for driving it make forwards relative to a heading rather than a spot on the vehicle. You can then drive forward, back, or side to side relative to that heading. You could then add control to rotate the tank, as well as a control to rotate the heading. I think this would make it a lot easier for trying to twist your way onto obstacles, as forward would always be facing over the obstacle, so you can twist the tank around relative to that heading without having to continually mentally readjust which way you want it to move.

This is fantastic 👏 👌 If you can set it up to run G-code and attach a wood router you'd have a cheaper version of the Goliath C N C Machine which costs about $8000.00 US Dollars. The low profile track isn't a mistake it just needs to be utilized in the correct project like a clone of Goliath. The router only needs to be mounted on a Z Axis Slide and the tracks controlled similarly to the Mazlow C N C Machine

"If there are only 4 directions why are there 6 pedals" in real life. Never thought I'd see the day.

You know what would be great: make it tetrahedral!

New print tunes are an improvement

a life size version should be made to test how well it really performs as the basis for a drive train that we can ride on. Three treads could mean no need for insurance, driver's license, or other government interference in our lives.

These pivots most certainly need some spring or rubber powered suspension to get better traction upon lifting itself onto surfaces. I also wonder if it'd help any if the tracks had a slight \_/ shape with the wheels positioned in the corners. Now that could potentially become an odd shape and snag on things... But I frankly am not sure how to solve the back track perpendicularity issue. Lift the whole track up mechanically? 😅 Guess there's a reason we have omniwheels and not omnitracks - people haven't figured out how to solve all these little problems yet. Also I would suggest adding a directional sensor like a compass or a gyroscope, and correcting the drive vector to that direction rather than following a fixed forward configuration. It'd be much more intuitive to control, I think. And would also open up a possibility of a "spin and forward" maneuver.

This is an improvement, but to make it proper offroad capable, just use 4 tracks in 2 parallel lines, with the little wheels mounted at 45 degrees. This has the added benefit of allowing them to be bigger without increasing the bend radius of the track. Look up the Scorpion from Halo for some inspiration, although it wasn't omnidirectional.

How reliable would the tracks actually be in an all terrain scenario?

I like to know a little more about what motors you used for this.

Make a 34 little powered tracks and put them all perpendicular on the bigger track! And make three of them. That would be cool. Haha, keep making amazing videos!

Have you heard of "field centric control" ? It is a concept i first heard of in the first tech challenge. It makes use of an imu to allow for pushing the joystick away and the robot will always go away from you

I suggest adding springs to the articulated tracks so they are forced back down onto the surface when going over obstacles. You probably won't make a third version, but for future projects that's my 2¢.

Instead of painting and adding little wheels ,, you can just add small or average size metal balls,, it might solve the issue of getting stuck while climbing up the obstacles

Cool concept but would’ve never work irl for so many reason biggest one being maintenance

Nice build!... Could you encorporate the wave drive in your three tracks and have the wave drive moving a track made of balls on axles?

My only suggest is to go to 4 treads now. Raise the2 not in use. All 4 down for a fast spin. 2 down for traditional movement. Change which set is down for sideways movement.

Might be complicated to implement (or simple, I don’t know enough), but to aid the control issue you could sync the motion to the direction you’re holding the controller. Left would move the bot to your left, forward would be forward, etc, then you have that fancy spinning stick to rotate the vehicle. :D No more figuring out which direction it’s facing!

Imagine what would happen if this side moving track design was employed by modern military vehicles. It would allow for unprecedented manoeuvrability. Add side thrusters and you could rapidly side dodge.

Or 3 tracks pointing out like a star, but one able to rotate 90 degrees on the yaw?

What if the tiny wheels were little treads instead?

Maybe a method to tilt a track, so the inward tilts up and the outside tilts down, to help angle it over when it gets stuck as prevalently shown.

Good track design and Omni wheels design might not mix well. Most tracks use raised rubber ridged or grousers to give the track good traction on loose and uneven surfaces, but in this case adding that extra friction will work against the sliding that you want for it to be omnidirectional. You might be able to use the Omni wheels for this purpose though by making them stick out more, giving them a triangular profile and rubberizing them and/or making them a little bit soft so they conform a little to the surface

for connecting the belt to the 3d printed gear, how did you make the 3d printed one? did you know the pitch and stuff? how so?

would just 2 tracks turn better with the tpu wheels?

You could try a design where each track is rotated by 90 degrees. This should make traversing obstacles much easier. Designing the central hub will be more challenging tho.