GPS accuracy?

[mab2]

Hi, just a quick question for a project (that's just in my head ATM), but I wanted to build an autonomous vehicle to go around a field and was wondering how accurate GPS receivers are now.

Looking at the spec of a 66 channel, searching (22 tracking) GPS module, it claims 1.8m which would be fine (and a lot better than my old handheld Garmin GPS72), but I wondered how accurate it would be in the real world?

Does anyone have any experience of these modern GPS modules? Are they really that accurate? I was thinking that I could 'teach the autonomous vehicle the boundaries and then it could use the GPS to stay within those boundaries.

 

[misterT]

I don't know how good GPS receivers are today. Depends on the module.. cheap ones are less accurate of course. I know that forest harvesters can alert if the operator is about to cut a tree outside the given area.
To get really accurate location, you need a fixed GPS station at known location to give you the error produced by athmosphere etc. This is called Differential GPS (DGPS). Google it.
https://en.wikipedia.org/wiki/Differential_GPS

 

[RCinFLA]

Normal commerical GPS is about 3 to 5 meter accuracy with SA off on satellites. There are a number of ways to make GPS more accurate. Differential GPS is one way. Another way is to use the phase reference provided by military GPS L2 band 1227.60 MHZ signals to improve L1 commercial band accuracy. There are quite a few other ways to improve accuracy. Having good antenna and extended correlation peak can take accuracy to a few feet but it needs a very low noise oscillator reference in receiver.

Claiming a particular accuracy is a very slippery statement. The 66 channel receiver claiming 1.8 m accuracy is likely for optimum conditions in open area in a non-moving (vehicle) situation. If this can be counted on then 1.8 m will be achieved. Put the same receiver in a downtown urban canyon with multipath signals bounding all around off of buildings, driving at 40-50 kM/h and it will be considerably worse. If utilized properly having more channels and correlators in a receiver allows for reduction in multipath degradation and satellite coming in and out of range.

As many ways to improve GPS there are an order of magnitude more ways to have it made worse by reception environment. If you know these things and control them then a few feet is reliably achievable.

Trimble makes systems for autopiloting agricultural tractors around a field.

https://www.trimble.com/agriculture/autopilot.aspx

 

[edeca]

Selective availability has been discontinued and the US Government (who run the satellites) are committed to ensuring it cannot be used again. A useful point in GPS history.

More info here: https://www.gps.gov/systems/gps/modernization/sa/

 

[Diver300]

To get the best accuracy, a clear view of the sky is vital. That is why surveying GPS receivers have large aerials on a poles, to get them above the surveyor's head.

The tiny GPS receiver in a cellphone, in the user's pocket, is far less accurate, but usually good enough for vehicle navigation, or finding which building the person is in.

I don't think that the cost of the electronics makes a lot of difference. The GPS receivers will say how many satellites they are looking at, how good the signals are, and how accurate the position is likely to be. The more critical the application, the more accurate the data has to be before it is used. For your autonomous vehicle, if you position the receiver high up and unobstructed, and don't let the vehicle move unless the signal is good, you will have it escape the boundary less often.

 

[mab2]

OK, that fills in a few blanks .

The agricultural autopilot is sort of what I had in mind - though without paying for a commercial ready-made system.

basically, if I can get consistent positioning (so that a vehicle can cover a field without missing bits or overlapping areas already covered, too much) is what I'm looking for.

The autopilot claims 1 inch repeatability - that's impressive - I suspect it wouldn't be practical to do a diy equivalent for a 1 off, alas. I wonder if one of the differential systems is useable here...

Thx folks

m

 

[dougy83]

Combining GPS with other sensors can be used to increase the accuracy https://www.google.com.au/search?q=gyroscope+accelerometer+gps+accuracy
Perhaps of particular interest https://www.theregister.co.uk/2013/02/14/gps_boffins/ and same again here https://gizmodo.com/5984176/two-cheap-sensors-could-transform-gps-navigation

Because mobile phones have both GPS and accelerometers, you could use a mobile phone to control your automatic mower or whatever you're thinking of (an IOIO [https://github.com/ytai/ioio/wiki] could be used to connect the phone to the external controls of the vehicle). If you actually need the gyro as well, you could connect that to the interface board and read and process it with the phone.