Drive train idea

Anyone ever tried this? A four wheeled robot that uses Ackerman's steering, but has mecanum wheels, all powered. Would this be good, bad, what?

[lemiant]

A good thing to do when thinking about designs like this, ask yourself "what do I gain?"

[Jeremy Germita]

With mecanum drive, you have 3 degrees of freedom. X Translation, Y Translation, rotation
With ackerman drive, you have 2 degrees of freedom. Y translation, rotation about a certain point(determined by your wheel and structure geometry).

It seems like you will be losing a degree of freedom for what gain?

My 2 cents...
Jeremy

[Taylor]

Admittedly, I haven't done the math, but it seems to me that if mecanum wheels are not parallel to each other, they lose all their superpowers.

-edit- What if they're set at 45* angles relative to the frame a la killough so the rollers are in line with the robot?

[Ether]

Originally Posted by SuperNerd256

Anyone ever tried this? A four wheeled robot that uses Ackerman's steering, but has mecanum wheels, all powered. Would this be good, bad, what?

Ackermann is just a subset of 3DoF control.

Ok, thanks! We were mixing different wheels and drive trains at yesterday's meeting, and this seemed like a combination we yet to have seen. Our vision was a car-like robot that could swerve around corners and the straighten its wheels and move sideways. Just something for the off season.

[Andrew Schreiber]

Originally Posted by SuperNerd256

Ok, thanks! We were mixing different wheels and drive trains at yesterday's meeting, and this seemed like a combination we yet to have seen. Our vision was a car-like robot that could swerve around corners and the straighten its wheels and move sideways. Just something for the off season.

Sometimes there is a reason you haven't seen something done.

Consider looking to 2008, several interesting drive systems came out of teams that year. Of particular interest to you might be 65's ackerman and the 'twitch' drive system. (CDM search the tag 'twitch' and you should find it)

Part of my objection to a system such as you described would be complexity of code for little benefit. When mecanum wheels are fixed you already have 3 controllable degrees of freedom. Their behavior is very well known (thanks, in no small part, to Ether). The issue is that once you start rotating the wheels you add significant complexity into the calculations while not adding any real value (You don't gain additional DoF and, it could be argued, you actually lose capability due to additional weight and use of an additional motor).

[JamesCH95]

Originally Posted by SuperNerd256

Ok, thanks! We were mixing different wheels and drive trains at yesterday's meeting, and this seemed like a combination we yet to have seen. Our vision was a car-like robot that could swerve around corners and the straighten its wheels and move sideways. Just something for the off season.

If I understand you correctly, you could accomplish this with two pairs of steered wheels.

[Ether]

Originally Posted by Taylor

-edit- What if they're set at 45* angles relative to the frame a la killough so the rollers are in line with the robot?

Start with a standard mecanum wheel arrangement (Figure 1).

Swap the left&right rear wheels (Figure 2).

Toe-in the front wheels and toe-out the rear wheels (Figure 3).

Now you have a vehicle that can drive forward and backward (powered by rear wheels only), and has front-wheel steering, but cannot strafe effectively. Not an improvement.

[Racer26]

Originally Posted by Andrew Schreiber

Sometimes there is a reason you haven't seen something done.

Consider looking to 2008, several interesting drive systems came out of teams that year. Of particular interest to you might be 65's ackerman and the 'twitch' drive system. (CDM search the tag 'twitch' and you should find it)

Part of my objection to a system such as you described would be complexity of code for little benefit. When mecanum wheels are fixed you already have 3 controllable degrees of freedom. Their behavior is very well known (thanks, in no small part, to Ether). The issue is that once you start rotating the wheels you add significant complexity into the calculations while not adding any real value (You don't gain additional DoF and, it could be argued, you actually lose capability due to additional weight and use of an additional motor).

Twitch came from team 1565 in 2008. It was called the linkage drive, and it was a 4 omni-wheeled robot, and the linkage would switch the wheels from pointing N/S to E/W. Switching at speed resulted in significant coasting in the direction of initial travel, creating arced movements around the racetrack.

Look for pictures from Waterloo or Greater Toronto to see it. 1565 is currently defunct, having not competed in the 2011 season.

Originally Posted by JamesCH95

If I understand you correctly, you could accomplish this with two pairs of steered wheels.

But it wouldn't be as cool!

[JamesCH95]

Originally Posted by SuperNerd256

But it wouldn't be as heavy or complicated!

FTFY