We use either tracks or two We use either tracks or two wheels + some sort of support because it makes it easier to turn. If you build your robot with 4 wheels, like a car, you have to make the front wheels turn, like a car and add a differential to the rear wheels because in a turn the rear wheels will not rotate with the same speed (see http://en.wikipedia.org/wiki/Differential_(mechanical_device)). Using two wheels or tracks, turning is as easy as stopping one motor or running the motors in opposite directions. It is mechanically more simple to use the two wheels and ball roller as opposed to add a differential gear and car-like steering.
Has anyone built or seen info about self-steering four wheel robots? The mechanics don’t seem to be too challenging - RC cars do it. I’m just not sure how complex the programming will be.
I’m also wondering about driving the steered wheels - like many cheaper cars - rather than separating out driving and steering. The mechanics gets more complex, but the logic may be simpler.
I think people use 2 wheels I think people use 2 wheels because they were just follow exactly as Fritls instruction shows. I just tare my robot apart and make it track. Track are so cool. I’ll never use wheel again. I am in love with track.
That said, Frits DID produce a four wheel drive vehicle with differential drive which steered by bending in the middle. I think this is an excellent control idea, but would benfit from a single engine driving all four wheels with three differentials.
All of the primary methods are fairly similar. 2 wheels plus a caster (known as differential drive) can turn in place. Tracks can turn in place. 4 wheels (each side has to be driven independently) can also turn in place, its just like tracks – with one caveat – you can’t have high grip/friction tires, or the thing will jump all over the place.
If your robot gets more advanced, and wants to know where it actually is, it will typically need the addition of encoders on the drive wheels to measure movement. This is where the 2 wheels and a caster wins. The other methods skid-steer, and so the errors they generate pretty much ensure that the encoder outputs are highly inaccurate compared to the real movement. Given high friction/grip tires, differential drive can be very accurate for closed-loop control.
Tracks look cool - it’s that Tonka thing. I have a feeling that they are not very efficient.
You need high grip in the front/back direction to ensure that the tracks don’t slip over the ground, but you need low sideways grip to allow the robot to skid steer. It’s going to be harder to use shaft encoders to determine position because of the cumulative erros due to slack in the tracks (if there was no slack the robot couldn’t move).