Understanding RTK and NTRIP for drone surveying and mapping

[00:00:02] Interviewer: Your Engineers have to have a good understanding of these things. Why is it so important?

Bob Foley: NTRIP and RTK. So the problem we're trying to solve with these is we want to position something in 3D space, okay, we want to position something.

More precisely, what we're trying to position is our drone, but actually what we really want to know the position of is our camera. The sensor of our camera. Exactly where the sensor is and what it's pointing at, we want to know that as precisely as possible. For photogrammetry, one of the key elements to understand is the position of the camera relative to all other images, and also the absolute position of the camera in the universe or in, on Earth really more accurately.

So we want to know exactly where that camera is and more precisely, exactly where the sensor of that camera is, because if we know that using photogrammetry we can calculate a lot of positional data within the images it takes. So if I know that a camera is 20 metres above an object facing straight down, what is called nadir. And I take a picture with a certain type of lens, with that image alone I'm able to make calculations about what is in that image.

I can say that the wall over here is this long and I can say that the house over here is this high and things like that. And once I start to get multiple images, now I've got 2 different images of the same thing from two different positions I'm able to make 3 dimensional calculations on pixels, or what we would term objects within those images. But the key element to understand is that you need to know the position of the camera when you take those images, it needs to be as precise as possible.

So how do we attain the position of a camera? For photogrammetry or LiDAR or anything like that, which needs to know its position, we use GPS. Okay, so we use GPS. So GPS all the satellites flying around the earth and they're all sending out a signal, each one sending out a signal that can be interpreted by all sorts of devices to give you your position.

[00:02:26] Standard everyday vanilla GPS will give you accuracy of maybe 20 to 50 centimetres depending. Maybe up to a metre, but that's the sort of accuracy you can expect from everyday vanilla GPS. It's more than sufficient, for example, to be able to drive a car down a road and for Google Maps to know that now I've turned left, or I've gone onto a slip road or something like that. But it's certainly not accurate enough for us to be able to map down to one and two centimetre levels of accuracy.

That's not possible. So we need a way of making GPS much better. GPS works on the principle of what's called Rho, Rho navigation - Rho being the Greek symbol. What it means is - if I know there's a fixed object and I know that I am 20 metres or 20,000 kilometres, it doesn't really matter, from that object I know I must be, if I'm on Earth, on a circle exactly 20,000 kilometres radius from this fixed object.

I don't know where I am on that circle, but I know I'm on that circle. Now if I have another fixed object. And I know, I'm 10,000 kilometres from this fixed object, now I know that I can only be in two places where the two circles interact with each other. Yeah, yeah.

If you ever see a Venn diagram with circles, you'll know exactly what I'm talking about. The radius, the circumference of the circles only interact at 2 points. I add a third circle. Now I have three points, alright now I can only be in one location. So just by using 3 satellites or three objects, known position objects and knowing my distance from them, I'm able to get that information.

So GPS works off that principle. It's called the principle of Rho, Rho navigation, yeah, where we know I'm a distance from an object. And I know, I know I'm this distance from this satellite, this distance from this one. I put them all together, I'm able to get my position.

[00:04:39] Instead of circles with satellites, we use spheres, as in, we are on a sphere exactly distant from the satellite, the point is - how we figure out where we are is the distance from it.

In satellites, in GPS satellites, that distance is a couple 100 kilometres 20,000 kilometres. It's a big distance. The bigger the distance, the more there is for variation, the more chance there is for variation and inaccuracy.

Okay, so when we're dealing with a distance between the reference object - the satellite, and us, and it is a couple 100 kilometres 20,000 kilometres our variation can be large. And in the instance of normal vanilla GPS it is 20, 50 centimetres, 1 metre, maybe even 5 or 10 metres depending on where we are in the world. How do we improve this? We can't bring the satellites closer and we can't go to them.

That's not going to work. By the way, this distance between the reference object, in this case the satellite, and us, this is what's called the baseline. Okay, this is called the baseline. How do we improve that? Instead of relying solely on these satellites, couple 100 kilometres20,000 kilometres in space flying around in circles.

Now what we do is we also bring in terrestrial base stations. So in Ireland there's maybe 30 or 40 of these base stations and they act roughly like a satellite from our perspective, except instead of the baseline being a couple of 100 kilometres 20,000 kilometres, there's one in Cork city right now. You're in Cork city right now, so your baseline to that particular object might be only three kilometres, 2 kilometres or if I'm out in the countryside it might be 20 kilometres.

The point is the baseline is much smaller and therefore now our room for error is much smaller. Also these base stations, these terrestrial base stations are put in very precise known locations. The exact height and horizontal reference of these base stations is mapped out and put into the unit.

[00:06:49] It knows exactly where it is and for that reason you're able to decide much more accurately where you are relative to that. By combining the satellite network and then also what are called local terrestrial base stations, you now suddenly have a much better idea of what is happening.

NTRIP and RTK, what we're actually talking about is integrating things like these base stations, these NTRIP network base stations, into our positional calculations to improve the accuracy of the thing we're trying to position in 3d space.