Chief Delphi

Chief Delphi (http://www.chiefdelphi.com/forums/index.php)
-   FIRST Tech Challenge (http://www.chiefdelphi.com/forums/forumdisplay.php?f=146)
-   -   omnidrive vs. kiwi (http://www.chiefdelphi.com/forums/showthread.php?t=47204)

slent thndr 05-05-2006 20:01

omnidrive vs. kiwi
 
I have been surfing around looking at lots of omnidrives (4 omni wheels) and kiwi drives (3 omni wheels), and I am wondering which of the two is better?

The omnidrives would probably have more available power, but perhaps more friction as well?

Also, how do they drive? With the example of an omnidrive, do all 4 wheels spin during straight motion? Or do just the two oposite each other turn, and the other two coast?

I would experiment, but at the moment i only own 2 omni wheels :rolleyes: .

Dan Petrovic 05-05-2006 21:03

Re: omnidrive vs. kiwi
 
I'd rather use 4 omni-wheels. It would take more weight and all that but a rectangular chassis is easier to fabricate and deal with than a triangular chassis.

http://www.chiefdelphi.com/forums/sh...?postid=420541

To drive forward, all wheels drive forward. To go to the side, the wheels on the same side will drive in opposite direction. wheels in the opposite angles of eachother will drive in the same direction. To go left, the front right and back left would drive forwards and the front left and back right would drive backwards. This would be opposite to go right.

I'd expect that if you were going in a direction parallel to a pair of wheels ie. Front left and back right, then the other two wheels would turn at all.

I hope I didn't make that too confusing

slent thndr 05-05-2006 22:54

Re: omnidrive vs. kiwi
 
With the system you describe it seems that the robot is never driving in a direction which is perpendicular to one of the wheels, so why are the omni-directional wheels necessary? Unless you were traveling straight in a diagonal I guess... in that case would two of the wheels be turned off? For instance, if traveling diagonally forward and left, wouldn't front left and back right be off, and front right and back left be traveling forward?

Dan Petrovic 05-05-2006 23:11

Re: omnidrive vs. kiwi
 
Quote:

Originally Posted by slent thndr
With the system you describe it seems that the robot is never driving in a direction which is perpendicular to one of the wheels, so why are the omni-directional wheels necessary? Unless you were traveling straight in a diagonal I guess... in that case would two of the wheels be turned off? For instance, if traveling diagonally forward and left, wouldn't front left and back right be off, and front right and back left be traveling forward?

You are correct. The reason for omni-wheels is because you want less friction. When making an omni-directional drive, except maybe swerve/crab drive, friction is the enemy. It would work with the kit wheels but I can't imagine it would work as well.

I don't really think the omni-wheels are necessary, just strongly suggested.

slent thndr 06-05-2006 11:11

Re: omnidrive vs. kiwi
 
That sounds like it would be a nightmare to program. What is the typical way that the sticks control it anyway? For instance, does one control stick control straight, sideways, and diagonal movement, and the other stick (ch 4) control spinning? (that would be like a video game) Or would the right stick control forwards, backwards, and spinning like a normal bot, and the left stick (ch 4) control strafing (sideways motion)?

With the last mentioned approach one could possible leave the right stick in the normal 4 wheel drive/ 12 configuration, and just program ch 4 to mix in strafing (sideways motion).

Does anyone have an existing program for this that they could post?

My concern with the whole omnidrive idea is that a lot of the power is being wasted. Comparing forward motion, how much slower is an omnidrive bot compared to a conventional drive train design with otherwise the same weight and power?

Madison 06-05-2006 12:32

Re: omnidrive vs. kiwi
 
Quote:

Originally Posted by InfernoX14
I'd rather use 4 omni-wheels. It would take more weight and all that but a rectangular chassis is easier to fabricate and deal with than a triangular chassis.

http://www.chiefdelphi.com/forums/sh...?postid=420541

To drive forward, all wheels drive forward. To go to the side, the wheels on the same side will drive in opposite direction. wheels in the opposite angles of eachother will drive in the same direction. To go left, the front right and back left would drive forwards and the front left and back right would drive backwards. This would be opposite to go right.

I'd expect that if you were going in a direction parallel to a pair of wheels ie. Front left and back right, then the other two wheels would turn at all.

I hope I didn't make that too confusing

Another more likely possibility than that shown in the linked thread is a drive which rotates the orientation of the omniwheels 45*. Thus, to drive forward, two of the wheels are powered and two are idled. To strafe, the opposite wheels are engaged. For any other direction, some combination of the two is required.

Such a robot will travel forward and sideways at speeds comparable to other robots based on its gearing, though optimally, I think you'd like to gear it faster than other machines in those directions so it's not unusable when traveling at some other angle. When you're using a combination of wheels to move at some random angle, your speed will be something less than it's straight line speed.

Where omnidirectional movement is concerned, "forward" becomes a relative term. You can make forward on the joystick any direction, really. Some are more efficient than others, however, when speed and power are considered.

slent thndr 06-05-2006 13:52

Re: omnidrive vs. kiwi
 
The drive train you discribe (in which 'forward' motion involves 2 omni wheels oposite each other spinning, and the other 2 ideling) could possibly be slower/weaker than the one in the link, because in forward motion only the power from 2 of the motors is being harnessed.

Madison 06-05-2006 14:31

Re: omnidrive vs. kiwi
 
Quote:

Originally Posted by slent thndr
The drive train you discribe (in which 'forward' motion involves 2 omni wheels oposite each other spinning, and the other 2 ideling) could possibly be slower/weaker than the one in the link, because in forward motion only the power from 2 of the motors is being harnessed.

As shown in the link above, only a fraction of power from each of four motors is being used. It's six of one and half a dozen of another. As far as I'm aware - and it's early yet so I may not be thinking straight - you'll never get more tractive force in the direction of travel than from what you'd get on a regular 2WD robot. The motors from which that power comes and in what ratio shift as your direction changes, but the cumulative output will remain the same.

slent thndr 06-05-2006 15:49

Re: omnidrive vs. kiwi
 
what do you mean by 'six of one and half a dozen of another'?

billbo911 06-05-2006 16:21

Re: omnidrive vs. kiwi
 
Quote:

Originally Posted by slent thndr
what do you mean by 'Cand half a dozen of another'?

It's a commonly used expression. Depending on the context it is used in it could take on a few different meanings, such as:
"It doesn't really matter", "It's your choice","Choose either way", "Neither choice is better", "Similar results with either". Get the picture?
"Six of one"=6
"Half a dozen"=6
6=6

slent thndr 06-05-2006 16:37

Re: omnidrive vs. kiwi
 
Quote:

Originally Posted by billbo911
It's a commonly used expression. Depending on the context it is used in it could take on a few different meanings, such as:
"It doesn't really matter", "It's your choice","Choose either way", "Neither choice is better", "Similar results with either". Get the picture?
"Six of one"=6
"Half a dozen"=6
6=6

Oh, I see. I thought it referred to some technical equation that I don’t understand :D

slent thndr 06-05-2006 22:15

Re: omnidrive vs. kiwi
 
does anyone have a program to control a 4 wheel omnidrive that they can share?

Lil' Lavery 06-05-2006 22:56

Re: omnidrive vs. kiwi
 
Just a quick note first. Both 3 wheeled, and 4 wheeled systems are "omni-drives" (which also includes swerve ("crab") drives, mecanum drives, ball drives, and other omni-driections systems). The method of motion is referred to as a "holonomic" drive system, where the motion of the robot is based off of the vectors of the output of the motors. 3-wheeled holonomic systems are also referred to as kilo-drives, and kiwi-drives.

When discussing the 4-wheeled system, let's nickname each of the 4 motors/wheels North, South, East and West (assuming you are placing them in the cardinal directions, as opposed to the corners as shown in the picture, but the same principles still apply in either positioning). To go directly North, you power the East and West motors, but the N and S stay motionless. To go East, you power North and South, and E and W stay motionless. To go along a diagonal, you power all 4 motors equally. To travel anywhere in between, you power all 4 motors varying amounts. To get a curved path instead of a straight path of travel, you power one motor more than that on the other side. Say you want to travel northwards, with a slight curve to the west. You leave N and S idle, power East 100%, and West 90% (so it will curve like a skid steering, aka tank drive, system would). To spin in a 4 wheel system, you move two opposite facing motors in opposite directions, instead of the same (or all 4 if you wish to spin in place).

Likewise in the 3-wheeled system. To travel in the exact direction of one of the wheels, you leave that wheel idle, and power the other two motors equally. To travel in other directions and curves, treat it basically the same way as a 4-wheeled system, just the amount of power to each wheel will be differnt based on direction.

The difference in between systems is in the efficiency. In a 4-wheeled system, you get 50% efficiency from your motors. Think of it when moving along a cardinal axis. You have 2 motors powered, and the other 2 idle, therefore 50% efficiency (the same applies for any other direction, as some of the power is "wasted" as you are not travelling directly into it's output direction). In a 3-wheeled system, you draw 66.7% efficiency. You get 100% efficiency from both systems when spinning in place.
.667x3motors=effective power of 2 motors
.5x4 motors=effective power of 2 motors
Therefore, unless your spinning, you get the same power with 3 motors as you do with 4, but you use 1 less motor.
But with a 4 wheeled system, you have a greater "footprint" or support base, making you more stable. You also get a greater power output when spinning in place, although the value of that tends to be limited. The math to figure out the vectors (and therefore the programming) is also slightly easier though. As well as construction in a kit with all 90 degree angles, such as the vex kit (60 deg angles are obnoxious to bend).


As for control, with the vex controller, i'd recommend the following. Use the left (or right, depends on your preferance) to control direction of travel. Up, the bot moves "north". Left, it travels "west". Right="East", down="south". The other stick would control spin. Left=counter-clockwise, right=clockwise.

I'd leave out the curves for now, it's more difficult to program and to control, and they tend not to be very necessary.

slent thndr 08-05-2006 15:26

Re: omnidrive vs. kiwi
 
Oh, the 4-wheel omni drive in this video functions differantly.

Quote:

Originally Posted by artdutra04

It has its wheels laid out as northwest, northeast, southwest, and southeast, and northward motion involves all 4 wheels spinning forward. Having the northwest and southeast motors on, and the northeast and southwest motors motionless would produce northeasternward motion.

I suppose that there are just two ways to set up a 4 wheel omnidrive?

EricH 08-05-2006 16:10

Re: omnidrive vs. kiwi
 
Quote:

Originally Posted by slent thndr
It has its wheels laid out as northwest, northeast, southwest, and southeast, and northward motion involves all 4 wheels spinning forward. Having the northwest and southeast motors on, and the northeast and southwest motors motionless would produce northeasternward motion.

I suppose that there are just two ways to set up a 4 wheel omnidrive?

Yes, there are only two ways (as long as you are using only standard omni wheels--if you go nonstandard, who knows how many ways there are).
As for the robot in the video: the control is similar, but: to go north, all four go one way, south, all four go the other way. East or west, two go one way and two go the other way. To turn, all the motors spin like they were trying to spin the same turret (assuming it's a circle laid on top).

slent thndr 08-05-2006 16:29

Re: omnidrive vs. kiwi
 
So which of the two ways is best? what are the trade offs? In the system in the video, all 4 motors can be powered for straight motion, so there is more raw power, but a lot of it is wasted because the wheels are at an angle to the way you want to be going.

Lil' Lavery 10-05-2006 22:16

Re: omnidrive vs. kiwi
 
Quote:

Originally Posted by slent thndr
Oh, the 4-wheel omni drive in this video functions differantly.



It has its wheels laid out as northwest, northeast, southwest, and southeast, and northward motion involves all 4 wheels spinning forward. Having the northwest and southeast motors on, and the northeast and southwest motors motionless would produce northeasternward motion.

I suppose that there are just two ways to set up a 4 wheel omnidrive?

It actually functions exactly the same. Just imagine that one of the corners of the bot is north, instead of a flat face. I just used a motor to represent "North" for ease of explanation.
And you can orient the wheels in any fashion you want. It doesn't have to be directly aligned with a "cardinal direction", or 45 degrees to it. Your vector equation to control it just changes whenever you move the wheels' orientation though.

Quote:

Originally Posted by slent thndr
So which of the two ways is best? what are the trade offs? In the system in the video, all 4 motors can be powered for straight motion, so there is more raw power, but a lot of it is wasted because the wheels are at an angle to the way you want to be going.

As I mentioned in an earlier post, unless spinning (where the more motors does equal more power), both systems result with an effective power of 2 motors. 66.7% efficient 3 wheel drive, and a 50% 4wd. So, the 3 wheeled acheives the same power for most applications with lesser motors, wheels, and weight required. The 4 wheels grants a larger conservative support polygon, or "footprint" and you are therefore more stable. Additionally, a 4 wheel drive only requires right angles, while a 3wd takes 60 degree angles, which are much harder to contruct acurately with a vex kit.


All times are GMT -5. The time now is 20:43.

Powered by vBulletin® Version 3.6.4
Copyright ©2000 - 2017, Jelsoft Enterprises Ltd.
Copyright © Chief Delphi