localisation

Hello, I have three small Automated Guided Vehicle wich navigate in a 700*700 square feet area. Using à PC, i want to know at each moment where every one of these robots is. What are the different method i can use wich give precise results. Thanks

I would guess radio beacons would give you all kinds of problems in near field. If it's an open area with clear sight lines, you might want to try three LED beacons in the corners (modulating at discrete and easily filterable frequencies), and put a rotating sensor with a lens in front of it on each vehicle, along with the intelligence to do its own triangulation of the identifiable light sources or at least report angular position of the three beacons. The vehicles could then transmit each in turn to your PC, by radio link if you prefer.

Half-stepping a good stepper motor for the sensor should get you within better than a degree for the angle on each of the three beacons, which will give you an accuracy of within a couple of feet in the x and y axis from any position. Also, by measuring signal stength at adjacent steps, you might be able to improve the accuracy just a bit.

This is the type of problem that is very amenable to large infusions of cash to achieve better results. But if you've got some experience in electronics as well as robotics, it is doable.

Good luck Chris

one

field.

LED

three

radio

better

you

position.

cash to

as

In that size area a RFID tag on each plus three transevers might work . It depends on if the area is empty or full of walls also.

Charles.

Maybe GPS will work:

You might use a wireless 802.11 link between the handheld and your desktop. It won't give you totally precise values, but it should be relatively accurate to a meter or so. If you get a 4th GPS unit, you might be able to use differential GPS to locate the roving systems exactly.

Regards, Bob Monsen

I'd like to throw my three pennyworth into the pot.

Would it be possible to lay a matrix of wires on the floor area and give each robot a unique coded frequency fed to a coil in the base.

The X and Y wires could be interrogated for a pulse and the frequency would denote which robot was over a junction.

It's just an idea, someone may be able to enlarge on.

Kind regards, Gordon.

----------------------------------------------------------------------------

one

Yes, the area contain many walls.

With walls light based systems aren't really usable. So use some form of RF. RIFD tags are smallish and already exist. Using a tranmitter on each unit requires only that they have different frequencies and that the signal proccessing used be differential . By comparing each receivers signal strength against the other two you can form a location even if the transmitter is weaker with time.More receivers will also give improved location.

Charles

Tile the floor area in a checkered grid like this:

- alternating black and white squares, but each black square has a thin white stripe down the centre, thinner than the floor area of the robot, and each white square has a black stripe. The stripes are all facing north-south.

- Add in a series of barcoded north/south and east/west stripes every so-many grid squares, which give an absolute reference, in case of counting errors.

- have an array of reflectivity sensors on the bottom of the robot, which looks at the floor and works out where it is in the grid pattern by counting transitions from one grid square to the next.

you should be able to work out some algorithm based on this that will work out where the robot currently is.

Or tile the ceiling area. Just like in horry movies, people forget to look up!

position.

And you can certainly do more than just half step, for example, the Linistepper controller supports up to 18th over stepping or 3600 steps per revolution from a standard 200 motor.

Either accurate Radio Direction finding with triangulation or have each have a map of known obsticales and Ultrasonic detectors.

See the links here

one