GPS at FRC Regionals?

[seg9585]

Has FRC considered implementing a differential GPS system at future regional and championship competitions? It would add an exciting new automation option for teams, and be pretty educational when it comes to training students about today's approaches in UAV navigation.
I was hoping this or upcoming years we'd start to see GPS made available in the game (via repeaters installed within the arenas).

[BBray_T1296]

with the GDC, nothing is impossible. (even a water game )

Are you talking about real satellite GPS, or some sort of pseudo-"field-relative" GPS?
If the former, I would say no, for logistic reasons
-it would be very hard for the field set-up crew to position the field at EXACTLY the right spot, or else (even though your code was made right) you might end up 3 (or 30) feet off.

If the latter, I would say: "not in the immediate future"
-Don't get me wrong, it would be really cool
-it would probably be expensive for teams to have said system on their practice fields to prepare autonomous (whether FIRST gives them one or they have to buy one)
-Imagine Championships. With 4 fields (and 4 systems) running at once, what a nightmare!

For either:
-with the vast array of venues out there (from coliseums to high school gymnasiums), the system might act differently
-don't most large regionals already have trouble with so much radio activity?


You could always go with a gyro/accelerometer approach, rather than traditional "dead reckoning" or some sort of camera wielding AI

[ezygmont708]

RFID waypoints could be interesting...

[DonRotolo]

Differential GPS uses a local GPS transmitter whose exact location is known (and not moving). The local transmitter then compares its actual location with that from the satellites and transmits a correction fcator. This is effective over several hundred meters to perhaps a few miles (with degrading accuracy over distance).

The main issue I see is that even the best differential GPS is only good to about a foot in a dynamic environment (something that is moving). Yes, surveyors can get fractions of an inch, but only with observations over relatively long periods of time. That's not really good enough I think. Plus, the issue of receiving GPS signals inside a building add significant challenges.

Inertial navigation with external checks seems to be the ideal FRC navigational method. Inertial navigation tries to measure where you are, and the checks (such as ultrasonic ranging) help keep things real.

[Bpk9p4]

As for the accuracy of GPS you can get around 5 mm accuracy in a dynamic environment with a GPS base station. However these systems are very expensive.

Another major problem is the software needed to correctly interpret GPS information. To do this you need a Kalman filter and a good IMU. This is a very fun project but is well beyond high school level programming.

This all being said i would love to see FIRST implement a GPS system

[Alan Anderson]

GPS repeaters don't work the way you'd want them to. Any receiver getting signals from a repeater will just end up with the specific location of the repeater's sky-pointing antenna.

There have been occasional attempts to come up with a Field Positioning System using ultrasonic beacons. I tried to devise one myself a few years ago, based on the same idea as LORAN. The math got a lot more complicated than I expected.

[protoserge]

I think the biggest problem would be acquiring the GPS signal in the first place. Many regionals have been held in non-GPS-friendly facilities. Without running cable to an antenna on the roof, it would be unlikely to get a consistent signal.

[EricH]

I'd rather see 1-2 April Tags on the field. A team who knew what they were doing and the location of the tag(s) could get a pretty good estimate of where they were even if camera(s) on the robot could only see one.

An April Tag looks like a less-detailed QR code. I understand that by looking at size and angles of the lines in it, angle and distance can both be calculated.

[BBray_T1296]

Originally Posted by EricH

I'd rather see 1-2 April Tags on the field. A team who knew what they were doing and the location of the tag(s) could get a pretty good estimate of where they were even if camera(s) on the robot could only see one.

That is something we did with our vision targeting code in 2012. By knowing the exact size of the backboard, we could determine our distance (based on pixel counts) and had a system for interpreting trapezoids (ie: at an angle)

[seg9585]

Dead reckoning and double integration from accelerometers produce a large amount of error even during a 2-minute FRC round.
I like the idea of embedded RFID -- maybe they could be hidden in a grid configuration below the carpeting, and a sensor on the robot could update its position as it moves over the RFID.

By setting up multiple repeaters in an atneea you can derive your own geolocation, although at that point you may as well design your own field-relative system. I'm guessing you can get pretty accurate results by placing transmitters on the field boundaries.
You could possibly even accomplish it with a sonar-based transmission system -- derive relative position with respect to certain sound frequencies, all transmitted simultaneously from 4 corners of the field.

[wireties]

GPS is a line-of-sight technology. It pretty much does not work indoors.
Commercial GPS receivers do not give us the resolution necessary for FRC.
Differential GPS (like that provided by the Coast Guard for coastal navigation) is much better (maybe a few feet) but stiii not good enough for FRC. But some other radio-based navigation element would be cool in autonomous mode, perhaps a simple directional beacon of some sort?

[Al Skierkiewicz]

GPS has been tested by several teams in several venues with disappointing results. Many modern venues have too much steel in the roof deck to get a good GPS lock.

[seg9585]

Originally Posted by wireties

GPS is a line-of-sight technology. It pretty much does not work indoors.
Commercial GPS receivers do not give us the resolution necessary for FRC.
Differential GPS (like that provided by the Coast Guard for coastal navigation) is much better (maybe a few feet) but stiii not good enough for FRC. But some other radio-based navigation element would be cool in autonomous mode, perhaps a simple directional beacon of some sort?

GPS has been tested by several teams in several venues with disappointing results. Many modern venues have too much steel in the roof deck to get a good GPS lock.

As I described, adding multiple repeaters on the roof of the venue and re-transmitting the GPS signals indoors would resolve the indoor tracking issues. Differential GPS can provide inches, not feet of precision -- which is plenty of resolution for the FRC field.

Here's a quick article/caption that mentions the concept:
http://www.engadget.com/2007/02/21/u...your-position/

Apart from this, creating a local positioning system environment as described in this thread may be feasible using an RF based system (rather than attempting to detect field objects visually, which in my experience has been quite error prone due to unknown and varying field lighting conditions)

[Joe Ross]

Originally Posted by seg9585

Here's a quick article/caption that mentions the concept:
http://www.engadget.com/2007/02/21/u...your-position/

It's not nearly that easy. See this article for a description of the problem and one possible solution: http://www.insidegnss.com/node/1581

[lynca]

Originally Posted by EricH

I'd rather see 1-2 April Tags on the field. A team who knew what they were doing and the location of the tag(s) could get a pretty good estimate of where they were even if camera(s) on the robot could only see one.

Absolutely agree, I used April tags quite a bit to track robots and it was an excellent way to get position and orientation.