diving to 4 directions
 

I've heard of robots from the last compatitions that drove to 4 directions
(without turning I mean..) does any one know how?
thanks, Guy Rom

Re: diving to 4 directions
 

Hi Guy, Erev tov =]

There are actually a few different ways of driving to 4 diferent directions.

1) Omni Wheels:
explanation: http://www.chiefdelphi.com/media/papers/1836
picture: http://www.chiefdelphi.com/media/photos/26368
order at Andy Mark.

2) Holonomic drivetrain:
pic:http://www.chiefdelphi.com/media/photos/25310
info:http://www.chiefdelphi.com/media/papers/1836

order at AM

3) crab drive:
pic:http://www.chiefdelphi.com/media/photos/28618
info:explanation: http://www.chiefdelphi.com/media/papers/1836

those once are all custom made, fun fun fun!


If you want a team from Israel who used that stuff, you should probably contact the Neat-Team (#1943). They used a 2/4 omniwheels system in Rack and Roll. In addition, you should ask the same question at http://firstaid.forumcircle.com . as you might get there more answers from Israeli teams.

PS, what team are you from?

Re: diving to 4 directions
 

mecanum drive trains

Re: diving to 4 directions
 

There are several ways to achieve "omni-directional" motion (motion on both the x and y axis, as well as the ability to rotate). The most common methods are holonomic (and its subset mecanum) and swerve (crab). There of course are other methods (such as "ball drives" or hovercraft), but aren't typically very effective on the FIRST scale.
Holonomic Drives, sometimes referred to as killough drives or kiwi drives, are a vector-based drive system. They typically consist of 3 or 4 (typically) "omni-wheels" placed radially around the robot (although alternate configurations are possible with as few as 2 powered wheels). More specific math can be found in other posts, and in a link I will provide later, but essentially each wheel contributes a composite victor and the robot moves in the resultant vector (and is also effected by any torque created on the robot). The most common supplier of omni-wheels in FIRST is http://andymark.biz/
A subset of the holonomic that has become very popular in FIRST over the past few years (particularly in 2007 with the release of the AndyMark mecanum wheel) is Mecanum Drives. They function under the same idea as standard holonomic drives, but rather than have straight rollers and angeled wheels, they have angeled rollers and straight wheels.
Swerve Drives, also known as crab drives, are an omni-directional system in which the wheels themselves actually rotate to achieve motion in different directions. Wheels are typically built into "modules" which are rotated by one or more steering motors. In Co-axial swerves, all drive wheel modules are powered by a single transmission, but in most FIRST swerves each wheel is powered individually.

Since I have taken a considerable chunk of time to (distractedly) write this, I'm sure most of this information has already been covered. But you can get a more in-depth description of each, and the math involved by searching chief delphi. Consider searching for "omni-drives", "holonomic", "mecanum", "kiwi", or "swerve". Also, read this paper by Ian Mackenzie of 1114 dealing with the kinematics (math of motion) of omni-directional systems.

Re: diving to 4 directions
 

Sean lists pretty much everything you need to know. I have one minor nitpick to make though.

Traditionally, a true omnidirectional drivetrain allows translation in the x and y axes as well as rotation about it's center point, at the same time. A swerve drive will not allow you to do this. I don't believe a mechanum drive will either, but to be honest I haven't paid an ounce of attention to them beyond that cool Airtrax video.

Re: diving to 4 directions
 

You are correct that a swerve drive (at least any that have been produce in FIRST thusfar to the best of my knowledge) will not allow for translation and rotation simultaneously. Nor will it allow for near-instantaneous changes in the direction of motion (as the wheels have to be physically turned).
As for mecanum drives, it is quite possible to rotate and translate simultaneously. Mecanum drives function identical to their more "traditional" holonomic cousins, only that rather than having wheels angeled around the robot, the rollers are. I'm not sure how many teams have done this, but very few teams have achieved this same motion with holonomic drives either. It takes much more programming skill to achieve it, and a much greater use of sensors integrated into the drive.

Re: diving to 4 directions
 

A mecanum can theoretically rotate while translating. It also drives like a four-wheel with all the wheels independently powered. (In fact, driving like a four-wheel and only using the mecanum for short distances is possibly the best way to use one.)

Re: diving to 4 directions
 

A mecanum can REALLY translate and rotate at the same time...I've seen it in real life.

Theory becomes Reality....ah, the joy of robotics!

Re: diving to 4 directions
 

The definition of holonomic that I am familiar with is that all possible degrees of freedom must be controlled--but not necessarily at the same time. Thus, IMHO, Killough/Other omni-wheel designs/Some (67/111/469 style) Swerves/Mecanum = holonomic, 118 style swerves = omnidirectional.

Re: diving to 4 directions
 

thank you all, this really helps!
hope to see you all in atlanta ;)

Re: diving to 4 directions
 

In 2004 our team won awards for using an omni drive system, http://team1322.org/omni_drive.htm . Our system worked very well, we could go all directions including rotate. The trouble is finding students that can drive it. Their is too many directions avalable and then have to keep up with the game and points. In our system we solved the traction problem by adding rows of wheels. We could actual go fast and have a good amount of power.

Re: diving to 4 directions
 

A swerve drive with four independently steered modules, or one with two independently steered modules and casters, will allow for simultaneous translation and rotation. (If you do the vector calculations for each wheel, you'll quickly see that this scenario is very possible.) It's actually detailed in two of Ian's whitepapers. Some examples of FIRST robots who could achieve this are Team 47 in 1998 and 2001, as well as Team 1114 in 2004.