Hello,
I am creating a self-driving system that will be used in the field.
I want a GPS system that is going to give me cm accuracy, however, I want to do this without a base station.
I am looking at the “SparkFun GPS-RTK-SMA Breakout - ZED-F9P (Qwiic)”, however, I am not sure if I need another GPS unit with this to create a ‘moving base’, or if I can connect just to a GNSS network (am located in NZ) and gain cm accuracy with the readings coming from that?
I am new to the GPS world, so I am just wanting to get a good idea of how it all works.
Cheers
Welcome to the GPS/GNSS world! There is a lot to learn, like any new field, but it’s very doable with some time and effort.
I’ve had great success using SparkFun products both with a local base and using a network. I primarily use the Facet products in the US; I can’t speak to the specific board you are looking at, but they both use the same ZED-F9P processor and I would expect results to be similar. You’ll need a good GNSS antenna and, of course, solid network connectivity for your application. The ZED-F9P needs a decent (not perfect) view of the sky to get a good Fixed (cm) position solution. You’ll need to figure out, likely by experimentation, if the network available to you is sufficient for your needs. There are a number of approaches to streaming the network (RTCM) data feed to your GNSS receiver.
I’m a surveyor and I don’t work on self-driving systems, but my understanding is that “moving base” systems use a base & rover mounted on the vehicle to determine the vehicle’s orientation in space - that is determine a vector between the two units that are fixed to the vehicle. A fixed base system uses a base (network or your own base) fixed at known coordinates to determine the vehicle’s position in space.
For a moving vehicle, wouln’t you want both position and orientation? I believe that would require a fixed base on the ground (yours or network), a unit on the vehicle that acts as a rover for positioning and a base for orientation, and a second unit on the vehicle that acts like a rover for orientation.
Here’s a diagram from the uBlox user forum: https://portal.u-blox.com/s/question/0D … s-accuracy.
Tony.
I want a GPS system that is going to give me cm accuracy, however, I want to do this without a base station.
Problem is for RTK to work it needs an Anchoring Point from which to perform the relative geometry.
PointPerfect uses a signalling model derived from ground stations, but they don’t have service in NZ, and the accuracy is likely in the 6-10 cm range rather than the 1-2 cm that RTK
You’re not going to get sub-metre without some infrastructure or effort…
It might be helpful to imagine RTK as a triangle geometry problem, with the best fits of dozens of triangles, with distances measured in carrier wave-fronts constrained to sub-mm, from each of the satellites used, and thru time domain (so juggling these dozens of triangles, thousands of times), as everything in the system is moving.
The short side of the triangle is the 3D line vector between your rover and the base (notionally up to 10-20 km), and the long sides represent the different view points of each satellite some 20,000 to 25,000 km away (time of flight around 70ms)
You end up resolving two points in space, with high accuracy, and if one is fixed and known you can project that to the other one that is unknown, or you wish to establish.
There’s some 30+ years of academic material on this, so should be some material you can find / refer too.
You may have a look at PPP:
https://rtklibexplorer.wordpress.com/20 … solutions/
I’m doing some evaluations right now. You will need to stay static a long time under clear sky to get accuracy approaching centimetre. Once enough accuracy is reached, you will need to stay under clear sky without any signal lost/cut like driving under bridge or trees.