next topic »

[on:]

Hello all,

I would like to design an autonomous robot that can go to a point A, B or C in my apartment and then go back to its departure point.
I have seen the Roomba vacuum cleaner robot doing this with great precision (video) but I ignore how it is achieved.

My robot will have 2 directional wheel, similar to this example.
I have been thinking of mounting 2 rotary encoders on the motors shafts to "record" how far the robot has been and later take the same path backward but I'm not sure if that will be accurate enough.

If anyone have done such thing before or have any suggestion or advice for my project, I will greatly appreciate :-)

[Reply #1 on:]

I'm sure that recording wheel rotation won't be accurate enough. If you had a beacon at the 'home' position and your robot has a sensor that can find the direction of the beacon that would enable you to find 'home' once the robot was close, but getting close is still going to be a problem if it has to negotiate walls and furniture. It seems to me that getting a good robust solution would be very hard.

[Reply #2 on:]

How close is close? GPS is accurate to about six feet, correct? Maybe use GPS to get close enough to pick up the beacon? As for navigating around obstacles, the usual method seems to be some kind of range finder. If you could tie it to some kind of pathfinding algorithm (I've seen a number of them around), it should be able to get back to base.

Of course, once you've got GPS + beacon detection + range finder + pathfinding, you may be pushing the limits of what one Arduino can so (or blow right past them).

[Reply #3 on:]

GPS indoors might prove problematic. Usually, you need line-of-sight to multiple satellites. Even with quite clever differential GPS I don't think it'd tell you more than which apartment you were in. If it's going to be blindly stubling around trying not to bump into things, it seems to me that you need some sort of maze solving algorithm. I don't think it'd be feasible to do a complete map building solution in the memory available to a standard Arduino, but perhaps you can get a wall-following algorithm to work.

[Reply #4 on:]

Thank you all

I have been thinking about a GPS as well but I don't think it will be accurate enough especially indoor.
An accuracy of 50 cm would be great and yes, it will have a range sensor to avoid bumping into the furniture.

A beacon + a range finder + a good maze solving algorithm seems like the best solution.

About what sort of components would the beacon be made with? I have never seen anyone doing this before

Edit: I've a just found an awesome example.

[Reply #5 on:]

Yeah, further research on GPS sort of suggested that would be the case (since I'm going to be dealing with exactly the same issue soon).

Nice link you've found there, though. Looks like a good solution.