Dec 24, 2006

Is a GPS-sensor for the NXT possible?

From robots.net I caught this story about the World's Smallest GPS module, recently announced by Epson. The new chip is 7mm x 6mm x 1.28mm, so I wonder if that would leave enough room inside a typical NXT-sensor housing to fit the electronics required to have it interface with the NXT.

Does anyone know of any homebrew GPS sensors for the NXT?

What could you do with a GPS sensor on a robot?

Rob

12 comments:

msim said...

Well, as with most "the smallest in the world" titles, I would think that it would come with a high price. Not saying it wont be possible, because it could, but may be to expinsive for most of us.

Brian Davis said...

What could I do with a GPS sensors? Ah, if only... First,it appears it runs off a low enough voltage to be directly powered from the NXT, but the position resolution isn't perfect (10 m). However, this may be the absolute position accuracy... it might be possible to use this to do differential GPS to determine how far and in what direction the robot has moved with much higher resolution. I may end up being slightly bigger than that wonderful minimal footprint - it seems it needs a crystal oscillator, for instance. Still, very interesting...

Ideally, I' like this to move up to a LEGO-class entry for the SRS Robo-Magellan event:

http://www.robothon.org/robothon/challenge.php

But I can think of other uses, including robot cooperation (knowing where the other robts are is an important part of this... think "swarm mapping"), or tracking a runaway pet (or child). Originally I was thinking about using a BT-equiped GPS unit to communicate directly with the NXT, but that's hard under NXT-G due to the specific format of the NXT-G BT messages.

http://www.engadget.com/2006/12/08/emtacs-new-mini-s3-bluetooth-gps-receiver-is-smallest-yet/

(wow that's too long a URL...)

--
Brian Davis

Anonymous said...

With a GPS component available for NTX would allow you to build a system for navigation.

You feed into a NTX subroutine a list of GPS coordinates and have your NTX system navigate to the coordinate list. Using either a differential or 2-3 GPS receivers you can deal with the position resolution...and once you get to your approximate end point you can use the ultrasonic component to triangulate into the target.

I think it is great that a tiny GPS component is available!!

Daniel Wittenaar Brickbash Robotics said...

Well i think that a GPS reciever is nice for such applications. But lack of accuracy and a fairly HIGH price would be 1 bottleneck..

The other problem I think is how to process so much data. GPS is a constant stream of serial data. The NXT has no change to capture that info. He is just to busy with other things. 1 way to solve this is to make a microproc that handles the GPS reciever and put's some data into a memory that can be resolved by the NXT. Even than GPS hold lot's of data. If you only look to what a GPS location looks like: N52 19.059' E6 31.950'
This is a Waypoint. That is a lot of info if you compare it to the ouput of a simple digitale sensor that only ouputs a variable between 0 and 255.

And even than the NXT would have a large terain for GPS to work. Because GPS is not very accurate in a area 10x10 meter (30x30 feet) If a GPS reciever is accurate by 2 or 3 meter you have a very expensive reciever. Normal recievers for car GPS use have something like 10 meter accauracy.

Tony Naggs said...

Hi Rob

The chip size is not an important factor - this part and similar ones are currently shipping in a package of 4 times the volume.

The gateway question is what would use it for? An outdoors accuracy of 10m E-W and N-S, and much worse vertically, is good for a pilot or driver finding a landmark. Probably not fine enough to be useful on the scale of a typical Lego robot.

Of course you can try to configure a differential GPS setup. However Lego robots like to be in doors, away from puddles, big plants and the local wildlife. GPS does not like the indoors, presuming it can get a fix the flat surfaces of walls and floors can cause reflections and can mess up the accurate timings required.

Brian - these i.c. mostly does the signal processing. It certainly requires external components such as an antenna!


Apart from these little details a GPS receiver is straightforward. :-)

Happy Christmas all.

Tony N

Anonymous said...

Hi, I am trying with a GPS small module, GPS Holux GR-240 with Chaeron GPS Library, http://www.chaeron.com/gps.html#JavaLibrary

but you need a BT connection. I use Icommand 0.5.

When I have any result with This experiment, I will post the documentation.

JAB, Juan Antoino Breña Moral
http://www.juanantonio.info/

Brian Davis said...

First, wouldn't it be nice if Santa were bringing me a GPS-equiped NXT? The present that tracks itself... :-).

As to the cost of a GPS unit, yes, this isn't insignificant... but considering how much some of us are willing to spend on "toys", I don't think it's out of range either. The GPS add-on for the PSP is under $100 I think, which is practical for many of us. The datastream isn't too bad either - most GPS units can be "pinged", and only send information when requested. As to needing a lot of information to transmit, a Lat/Long set is just two numbers (and not very long numbers at that). Even if the GPS unit sends them as text strings, with all the charecters, that's just string parsing. Not a terminal problem by any means.

From the standpoint of differential GPS, yes, this is how I'd suspect it would be best used: thank of two seperate robots that "leapfrog", so that one is always stationary to provide a reference for the readings from the first (and with BT, having two robots coordinate is very simple).

--
Brian Davis
(who has watched GPS units with selective avalibility turned *on* still locate a point in space to within centimeters... differential GPS is fun)

Will George said...

Using either a differential or 2-3 GPS receivers.. GEEEE I'd need all the motors just to hall around the batteries to power it!

I'd think a good solution may be a Hall sensor 'looking' at a compass needle on the robot to detect North, south whatever.. AND go on from there.

Anonymous said...

Any discussion of the application of an active send-receive sensor should include the RFI sensor group. RFI taged objects are arriving on the scene (pun) very fast.

Anonymous said...

Someone already touched on this point, but without an antenna with visibility to the sky, a GPS is useless indoors.

Ben said...

Where can I get a GPS sensor for NXT??

Answer to All said...

Well First of all a GPS would be hard to communicate with unless it was done in another language like lejos and it would need a micro proccessor. But if you knew your lat and long and knew how much it changed in let say 1 rotation then you could have it done that way and have bluetooth running for swarms or whatever you what. That would need some more range which is eaiser. you could also use a camera sensor to read a GPS

Related Posts Plugin for WordPress, Blogger...