PDA

View Full Version : looking for high quality GPS



jwatte
04-29-2012, 02:46 AM
I'm looking for a high quality GPS. I'm currently using a USB BU-353, but I need significantly better precision outdoors, and would like better ability to pick up position indoors if available.
I have an ITX Intel board and a few homebrew AVRs on board, so USB, SPI, I2C and UART are all fine interfaces for this.

Slugman
04-29-2012, 07:56 PM
Re outside - Not sure if it helps, but we used to set up a base-station with another GPS on a known point, then adjust the mobile GPS to account for the variability in signal position as compared to the known position. Dunno if that method is obsolete since the military stopped inserting a variable into the GPS signal, but if your robot is controlled from a base-station anyway, then it may be something to look into. I.E. Don't rush out & get another GPS, use your favourite search engine to look into it first. :)

Re inside - Might have to triangulate somehow - Fudicials? I think there was info in the Trossen Tutorials area, but it's well outside my area of knowledge. Good luck!

tician
04-29-2012, 08:55 PM
Re outside - Not sure if it helps, but we used to set up a base-station with another GPS on a known point, then adjust the mobile GPS to account for the variability in signal position as compared to the known position. Dunno if that method is obsolete since the military stopped inserting a variable into the GPS signal, but if your robot is controlled from a base-station anyway, then it may be something to look into. I.E. Don't rush out & get another GPS, use your favourite search engine to look into it first. :)

Definitely not obsolete. It is called Differential GPS and is still used pretty heavily, with the FAA's Wide Area Augmentation System being a rather impressive example (remote stations measure variations in GPS signals and relay that data to master stations, which then send the data to geo-stationary satellites to relay the data to navigation systems).

For indoors, I'm not too sure there is any really good solution using GPS. Using DGPS and a bigger/active antenna might help, but no guarantees.

jwatte
04-30-2012, 07:16 PM
The intended use of this is for RoboMagellan, one rule-set of which I've seen explicitly disallow DGPS. Makes it too easy, I guess :-)
Reading up on this, there's also phase-based approaches that will do pretty good without base stations -- something like the PPP that's built into the u-blox 6 receivers. (Btw: where would I buy them? They're not widely available)

Slugman
04-30-2012, 10:17 PM
Sorry, can't help you with the receivers. :(
Just had a quick read of the rules. Looks like there's a lot of fun to be had designing the robot to do the course. :)
Just one thing you might think about re the GPS points - When surveying, if you close off on a point, you make a known point that can be referred back to with greater accuracy. Since they give you the location of each point, you could reset your location to the known point once you have touched or marked it off, so the only hard part is getting close enough to the next point to find it.

I.E. Use the GPS info they give you to create a virtual differential GPS by comparing your GPS reading with the known GPS point when your bumber switch activates when you run into the point. When you get to the next point, do it again! :)

Assuming it doesn't take too long between points & your GPS is relatively accurate (Set to averaging with a value useful for your robot speed) then the single "Differential" reading should improve things long enough to find the next point, or at least get close enough to find it another way.
A Roboard with a pan & tilt camera & Roborealm would be useful here, since you know the size & colour of the point/blob to look for. Good luck!

jwatte
05-01-2012, 05:28 PM
Thanks! That's actually similar to my current plan. Except I'm rolling my own based on an old large-scale RC hummer for the platform, and a couple of HD webcams and a mini-ITX motherboard for vision, and some perfboard AVRs for sensor and actuator/drive I/O. It's a lot of fun, which is the main purpose anyway ;-) I'm hoping to get some stereo vision based data out in the end, but I'm a far ways off from that yet.