Robot using GPS

"Brent" wrote in message news:WwjHb.41404$a44.15952@okepread04...

I have used GPS in a few robots, mostly water-going. With GPS, you are likely to only get about a meter or so accuracy in small movements over a short time, but the accuracy rapidly decays over any significant time to about 9 to 10 meters. For example, if you have a robot that is a meter across, and you place two small GPS modules on the longest part of the robot, as far apart as they can go, then they will likely measure a difference that you can detect. This is useful for determining the orientation of the robot without having to move around. GPS units typically look at your motion to determine your velocity and heading. This method allows you to take a static heading or orientation. And, if you are moving from a known reference location, you can use it to help calibrate your position quite nicely, but over a period of 20 minutes or so, you will have to recalibrate because new satellites coming over the horizon can greatly affect the apparent position of your GPS receiver. If, however, you place a GPS at a known point, and send its readings in realtime over, for instance, an RF modem, then you can subtract that reading from your onboard GPS unit(s) and get a far more accurate reading- this is differential GPS. Your robot will need an RF modem to receive the signals and your onboard processor will have to do the math. In other words, a single GPS module with no references is pretty crappy, but two GPS units can give you heading information, and a single GPS unit at a known position and broadcasting its reading to your robot (which also has a GPS) can be used to provide really excellent (within a meter) position reading, but you must invest a bit of construction and programming of your own to make it happen.

Cheers!

Chip Shults My robotics, space and CGI web page -

do you know of more accurate GPS devices? I hear of some being accurate to one millimeter but its unaffordable for any consumer.

Well, GPS is GPS- it is as accurate as it's going to get. In order to do better, you must use other references. That is the reason for differential GPS. It provides a known reference, and then subtracts the signal to use the differences. The reason that systems you might purchase for this are so expensive is that they use this method. More hardware, more measurements, a bit of software to glue it all together. The reasons for the inherent inaccuracies are based on physics and the hardware that was launched. But even the atmosphere contributes to the errors, as well as the fact that orbits are not perfect- they are affected by differences in density of the Earth itself (mascons) and even relativity itself causes some of the problems. Overall, when you get right down to it, the system is cheap and pretty good for most applications. When you start poking around with millimeters, you are far below the initial accuracy of the system and you must throw a few novel ideas at it to make it any better. Note that "novel ideas" = "cash" for most approximations.

Cheers!

Chip Shults My robotics, space and CGI web page -

I'm not real knowledgeable on the subject, but I remember reading that before the dither (sorrry, i don't know the technical term for this) was removed, people used GPS in conjunction with fixed known reference positions, to more accurately locate a position.