12-04-2006 04:16

sanddrag

Re: pic: Omnitrack Summer Project?

You're right. Making that one piece does seem like a job that could be done with a customextrusion. Sheet metal seems like a no-go if you ask me. You might be able to get the part lasercut or water-jetted out of thick plate, but it will have a draft angle, whigh might be okay though. Maybe you could put in some radiuses and have it CNC routed out of some good quality ply or some other good quality wood? Post on cnczone.com in the RFQ forum and people will jump at you with offers of how to make it and what it will cost.

I'm just curious, what is the problem with having it be two pieces of C channel as it is right now?

12-04-2006 07:34

Peter Matteson

Re: pic: Omnitrack Summer Project?

Why not band saw 2 different C-channels and rivet them together? That would basically do the same thing if you can get something close enough.

Also I would suggest solid aluminum rivets if you have the capability.

Edit: Missed that Sanddrag asked the same question.

12-04-2006 07:45

Raul

Re: pic: Omnitrack Summer Project?

Riveting the two C-channels together will be a weak link (no oun intended) in the design. Rivets are not meant to hold much axial force - for that you need screws. Even steel rivets will come loose as the tracks will see lots of forces in the axial direction of the rivets.

12-04-2006 08:46

Greg Needel

Re: pic: Omnitrack Summer Project?

First i would like to say i love this concept and see how it would be really beneficial. I do have a question about implementation though. How do you anticipate driving them?...the only thing i can come up with is a custom pitch sprocket that fits in between the links, but i am not sure that that will be reliable.

12-04-2006 08:58

Andy Baker

Re: pic: Omnitrack Summer Project?

There are some good points brought up here.

Riveting the two pieces together is not good.

Custom extrusions are expensive... however, there are MANY types of extrusions out there these days. This profile looks like something similar to what I see every morning when I open my shower door. Also, Paletti has an extrusion that is somewhat similar, SP106, but the roller size would need to change (and that would be a tragedy, I must say!). My advice regarding extrusions would be to keep looking. Someone may have a standard profile that may work for you.

Greg brought up a good point regarding how to drive this thing, and this point may re-define how your rollers are mounted. This would make your extrusion profile different, possibly.

I suggest starting the design with standard attachment chain and build up from there. If you can design a roller frame that can mount between attachment chain flanges, maybe one on each side of the chain, then you would have a double-wide omni track. If you can get this design to be based on a 35 pitch attachment chain, then you can use standard sprockets and drive transmission components.

Good luck, and please keep us posted of your results.

Andy B.

12-04-2006 10:07

Greg Perkins

Re: pic: Omnitrack Summer Project?

Wow, just looking at this design has got me thinking, thanks for the idea! Anyway, what I propose you do, is design your "sprocket" out of hex stock or custom octagonal stock. Use them as your "wheels" to drive them, then design your tracks with less links so they contact the flats on the sprockets. I did a really crude sketch, but there's the concept, seems (IMHO) the easiest way to drive it.

12-04-2006 10:15

Greg Needel

Re: pic: Omnitrack Summer Project?

Greg,

It is a good thought but remember that with your design the roller is the only thing that would be touching the drive. I guess if you inset the roller profile into the drive hex then it could work i just see long time issues with using the rollers as the driving feature in this assembly.

12-04-2006 10:20

Qbranch

Re: pic: Omnitrack Summer Project?

Hey if you need any help with machining, I can give you some advice at the least. Just send me a PM if you need any help.

I do CNC as well as manual machining.

-Q

12-04-2006 10:21

Greg Perkins

Re: pic: Omnitrack Summer Project?

I see what your sayin Greg, and trust me I know what you mean. I got some ideas up my sleeve, and when I get some time on my lunch break I'll actually start doing some CAD. It'll take some engineering, but you would have yourself something revoloutionary. Keep innovating!

12-04-2006 16:04

Bill_Hancoc

Re: pic: Omnitrack Summer Project?

I had an idea smiler to this a few weeks ago for a crab drive tank system. It involved 4 little individual triangular tread drives that could pivot on a dolly type system and then would be able to turn parallel to each other. The individual tread drives would be about 6-7 in long on the bottom and be based off a slightly modified 30-60-90 triangle. I wanted it to be fast along with really powerfull via a shifting gearbox but after calculating the speeds needed for a 1.5 in tread pully it would have been very hard to go fast since it would have reuired a really high rpm higher than the small CIM is i remember.

12-04-2006 16:32

RogerR

Re: pic: Omnitrack Summer Project?

in 2001 used belts instead of wheels to help them climb the midfield barrier. after a bit of searching, i found a pic here (thanks 229).

based on my calculations, to travel at 15 ft/s with a 1.5" driving pulley, you'd need almost a 2.5:1 gear reduction, which could be done in a single stage.

i think the biggest problem to overcome with your design would be turning the modules themselves; 6-7 in of high friction material would create a pretty good resistive moment.

this design is obviously do-able, there's just a few problems to solve; but then, how much fun would all this be without any problems?

12-04-2006 16:35

Simon Strauss

Re: pic: Omnitrack Summer Project?

I'm thinking that this would work better if the 4 treads were positioned like traditional wheels would be and the rollers at either 30deg or 45deg so that it would function just as mecanum wheels but with the advantages of treads.

12-04-2006 16:48

Greg Perkins

Re: pic: Omnitrack Summer Project?

I assume you mean Omni wheels. And no, this won't behave like tank treads, it's a whole different animal. You've got to take into account being pushed, your coefficient of friction, and the need of the design. Is it doable, yes...is it practical, don't know. Do some research, and let us all know please!

12-04-2006 17:09

Gabe

Re: pic: Omnitrack Summer Project?

You all have gone through my same thought process. I will address every point, since you all understand what the idea is:

The links are made from 2 different C-channels and each riveted with two 1/8" aluminum pop rivets. The problem is that this is very weak and time-consuming. I have not considered screws, and I will try it. What size: 4-40 or 6-32? Nylon insert locknuts, or tapped holes?

To drive the tracks I am hand filing 2 drive sprockets out of 1/8" aluminum sheet, distanced apart using standoffs. This sprocket engages the ends of the 3/16" steel pins.

I was thinking about the standard attachment chain, but its would be much heavier if I used 2 chains. However, I might scrounge around and find some other material that would work for small rollers traction rollers to be used on just one chain.

Many people have suggested using waterjets or CNC mills. Why does a waterjet or CNC mill seem right for this project? Do you fabricate a bar with the correct profile, or do you cut individual links?

I do have a mill: my 14" Nicholson mill file. Ha, ha! Yes, do file my own #25 and #35 sprockets out of aluminum, and yes, i Am CrAzY. No need to buy those IFI sprockets!

Sorry about that. Just a random fit of insanity.

12-04-2006 20:40

Cody Carey

Re: pic: Omnitrack Summer Project?

Why not CNC punch them out? I am fairly certain that you can
punch 1/4 aluminum, and it would be alot more consistant/pretty.

Also... Why omnitracks? Won't a 4-6 wheel omnidrive perform just as well as a two track two wheel design?

12-04-2006 21:46

Tristan Lall

Re: pic: Omnitrack Summer Project?

Extrusion requires a difficult setup and expensive tooling; it's only cheap in huge volumes. But what about graphite permanent mould casting? If you can get a local casting shop to sponsor you, it wouldn't take much to make a few hundred links.

Basically, you'd need to mill out the mould into two graphite halves with sprues, vents and risers as necessary (usually CNC—a casting shop will be able to do this easily), and it's just a matter of mounting it to a standard fixture (which opens and closes the halves) before you're ready to pour metal (e.g. 390 aluminum).

12-04-2006 22:34

devicenull

Re: pic: Omnitrack Summer Project?

So, assuming you are going with a standard tank setup, whats to stop someone from going on the side of you, and pushing you straight across the field?

Doesn't look like it would offer much resistance to that.

12-04-2006 22:48

thoughtful

Re: pic: Omnitrack Summer Project?

He explained it in his first post. If you can imagine, there are four treads, at each edge of the robot. So if someone pushes you from the side, the front/back treads should resist somewhat, in theory the system, if made, should be better at torque that the OMNI wheel drives.



Just a suggestion, instead of having 4 treads, why not have 2 treads on each side, and one in the middle(which intersects the turning center). This way if you need to implement the drive on a robot that herds or uses its back or front for other things, you will be able to fairly easily use the same concept. On a side note, this will make the robot more vulnerable to pushing.*Just a random thought*

12-04-2006 23:00

Jizvonius

Re: pic: Omnitrack Summer Project?

I think that Andy brought up a good point about using standard chain attachments with links mounted to them. I have been thinking about doing this for a little while, but haven't found the time. Maybe you can puts this to good use before I can.

Connecting links with mounting flanges can be found at McMaster on the last page of roller chain.


If you can get something similar to the part shown in the attached pic stamped out of plate to hold the rollers, I think that it will work when mounted to the chain. By properly sizing these parts, you may be able to make the chain lock in one direction giving the omnitrack rigidity between support rollers, but the ability to collapse around drive sprockets.



This is of course just concept, and needs tweaking.

12-04-2006 23:53

Simon Strauss

Re: pic: Omnitrack Summer Project?

No, i think i know what I'm saying as confusing as it may be, let me explain then feel free to correct away(i might want to pursue this idea of mine). It has already been established that this system is going to be implemented in four different treads each at a corner of the robot, my idea is to instead of having the rollers 90deg from the direction of the tread moving as shown in the picture, have them set at either a 45deg or 30deg and leave the treads parallel to the other side. My reasoning for believing that this may work is that if you imagine traditional tank treads as just an elongated and spoke-less wheel that travels in an ellipse to move its load instead of a circle, setting the rollers this way will function as nothing more than this perception applied toThese Mechanum wheels on the first page, with all the drive and other characteristics of the wheels applying to them. If this however is too confusing because my mind works in strange ways the treads will be positioned like
l l
l l
and the rollers positioned like
<>
just an idea and ill try to make a more detailed drawing to clarify what i mean when i get time
-Simon

13-04-2006 00:23

s_forbes

Re: pic: Omnitrack Summer Project?

I thought of omnidirectional treads too when trying to think of an interesting drive system and it's nice to see that someone else is persuing the idea.

One problem I thought of though is this:

If you have 1 tread on each side of the robot, then each tread will only receive about 25% of the total weight of the robot. When you try to push someone forward, you will only be using two of the treads. Since only 50% of your robot weight is on those treads, then you have a significantly reduced pushing power (less traction than a typical 4 wheel robot), even though you are using treads.

Then I remembered "Oh wait, omni treads would be sweet!" and it cancelled out the previous thought.

13-04-2006 02:29

Gabe

Re: pic: Omnitrack Summer Project?

True enough, and I have given much thought to different drive configurations.................................... ......................................... My favorite (4 treads):....Another idea (3 treads):...........Killough (3 treads):

== == ==.||....................||.............||........ ......................== == ==
||...........||....................||== == ==||..............................\..........//
||...........||....................||............| |................................\.......//
||== == ==................................................ .........................\..//

Realistically, these will take time and money to make, so I am thinking about this for a test platform (2 treads, 2 omniwheels working in tandem):

||.....||
||< >||
||.....||

To have the best performance on the field, you would have each tread independently controlled. Just to throw another idea onto the field, what if you used four DeWalt 3-speed shifters with FP motors, and had a 3-speed omnitrack robot? That would be incredible.

Why consider omnitreads at all? Two words: Freaky Mobility.

I am having problems attaching images. Periods represent blank spaces. Sorry.

19-04-2006 22:54

Tatsu

Re: pic: Omnitrack Summer Project?

With all honesty, this is cool, but it makes no sense to me.

You should put the wheels in a diamond pattern, so when you're pushing forward, you're at around 75 not 50% efficiency.

But what makes no sense, is why you want treads.
For non-deforming surfaces, frictional force is independent of area. IIRC, Tank treads make sense in that they can "dig in" and create normal force (instead of frictional force) but these AM rollers dont deform, dont dig in, dont really do anything but add weight. since there's only two points where the weight is supported (the aluminum sprockets) this is functionally equivalent to having two omnis on each side of the square.

Also, why are you filing? you have access to a SWEET cnc mill, it'd only require a fraction of the machining you did for your wheels

Tatsu out.

20-04-2006 10:43

Dillon Compton

Re: pic: Omnitrack Summer Project?

Uhm...If the rollers are at 45 degree angles to the tread, then the tread should be mounted parallel to the frame at the four corners; the nature of mecanum systems is such that you receive standard weight distro, speed, and forward traction, while gaining the manuverability of omniwheel systems.

Good luck.

20-04-2006 11:04

Tatsu

Re: pic: Omnitrack Summer Project?

I dont the the OP is talking about 45 deg rollers, and if OP did do that, he'd need the treads set up like traditional mecanum wheels (ie two treads / side) rather than the square shape he has now.

06-09-2006 21:10

Chuck Glick

Re: pic: Omnitrack Summer Project?

im just wondering how the work on this idea has been going along, because a couple of members on 1712 along with myself are working on a omni tread design as well. just wondering how far its gotten. and i think i figured a way to keep the treads in a straight line and have a good sprocket for them to move on.

07-09-2006 00:28

Gabe

Re: pic: Omnitrack Summer Project?

Not so well, because I realized that the tracks would weigh a lot if they were made of aluminum and if each had a 3/16" steel link pin. Also (heh) I noticed that with my design it is physically impossible to install the last link, and therefore it is impossible to complete one entire track. To put the 1/8" steel pin in, it would mean that it would have to go through the 3/16" pin.
If I were to change my design, I would make the links out of plastic, and have cylindrical rollers instead of barrel shaped rollers.

This will give you an idea of what I mean. Sorry about the image, but without AutoCAD, Inventor, or Solidworks, I have to work with what I have: Microsoft Word. This shows the top view of my idea, and you can see the cylindrical rollers with a pin going through (maybe 1/8"). The tread links are joined at the top and bottom by another pin ( maybe 1/8").