What's Better Than a Differential Drive Wheeled Robot?

[johnny555]

I know that sounds like the beginning of a super-nerdy joke, but I'm actually asking!

I've played with a few two-wheeled bots, setting them up as FPV video-streaming rovers (only on my local wi-fi network), but it gets frustrating that they rarely seem to drive straight. Can anyone suggest what my next move is - a four-wheel chassis? Or is there a jump up in tech (from the cheap-ish entry-level hobby platforms) that I should investigate?

Thanks!

[mklrobo]

  Hello!
In reference to your options, the vendors, here, that support this site, like servocity, have a
lot of different types of parts for the robot. I think that had a wheel that could turn left, without
turning the physical wheel left. Incredible! The wheel had parts in it that could be turned to
perform that task.
Insofaras your next jump in Tech, I would have fun looking over the goodies that the site vendors had,
while planning your next project for your bots. Looking on Youtube, what scientists and experimenters
have done, may give you inspiration to see if, "HEY, that's cool! Can I do that with my bot?!?"

[mklrobo]

  In addition,
These is a robot video section in this forum that may have some ideas for you,
posted by some robotic enthusiasts, like yourself. The trailer for Chappie,
 was cool.(but not related to your direction) 

[bdeuell]

It is possible the extra two wheels on a 4 wheel drive robot would help average variations out and improve your straight driving but i wouldn't count on it. The improvement would probably be minimal if any and its possible it could be worse.

Additionally turning will become very unpredictable because the four wheels the must slide sideways when turning. As the 4 wheeled robot turns in in a circle each of the wheels must follow a circular path. The rolling direction of all the wheels will never be tangent to the same circular path (unless the wheels are on the same axis), which will cause the wheels to slide sideways to some degree. This isn't so bad if your wheels are low friction but if you want your wheels to move your robot friction is a good thing. Also the wheels will tend to make and unmake contact with the ground (due to the sliding action and any uneven ground) which will cause the center of rotation to jump around. Imagine pushing your two wheeled robot sideways while trying to turn...not very predictable.

so what will help..... it depends on what you are working with.

Most likely if you have a low cost platform the wheels are not driving at the same speed. this is due to many sources of variation from the resistance of the motor winding to the diameter of the wheels. There are a few ways you can address this. You can calibrate your motors/controller to account for these differences but this isn't guaranteed to always work reliably. You could use stepper motors which are inherently position controlled motors. A very common solution is to use encoders to measure the wheel rotation and use feedback (PID) control to ensure both motors rotate the same amount.

If you are driving across uneven terrain (nothing is perfectly flat) each wheel will have a different distance to travel as it drives up and down the terrain. to correct for this you would need sensors to monitor your environment. A digital compass, gyroscope, or GPS would allow you to maintain a constant heading.

For complex robotic systems it is common to use several of these methods to robot both keep track of the robots location and ensure it moving as expected.

[johnny555]

Awesome - thanks, both bdeuell and mklrobo.

I like the idea of stepper motors. Was insanely impressed with their hyper-precise performance when I used a wall-mounted polargraph (specifically the Gocupi). The idea that the tiny movements of the two motors can draw a vector image accurately...it's kind of humbling!

Short of that, I suppose I'd need encoders and PID control; I'll have to do some studying up on that. Thanks again.