Igor is the CEO and cofounder of Emlid.com a company that started out making high-precision GNSS receivers in his kitchen and crowd-funded the first batch on Kickstarter.

But that was over ten years ago so today on the podcast Igor is going to tell us about the innovations that made this possible, give a great explanation of RTK and PPP and explain why we should expect to see high precision positioning on your phone any time soon.

Connect with Igor here: https://www.linkedin.com/in/igor-vereninov-52a73ab0/

Or visit https://emlid.com/

In this episode, we cover:

1. Introduction to Emlid and its Focus: Introduction to the company Emlid, its CEO and co-founder Igor, and their focus on high precision GNSS receivers and software designed for centimeter accuracy positioning.
2. Startup Story and Crowdfunding: The origin story of Emlid, starting from working with drones in university, the need for accurate maps, the initial challenges with high precision GPS technology, the development of their own GNSS receiver, and their successful crowdfunding campaign on Kickstarter.
3. Innovations in GNSS Technology: How Emlid managed to make high precision GNSS technology more affordable and accessible, the role of open-source software, and the technical innovations that allowed them to reduce the cost and size of GNSS receivers.
4. Market and Technology Evolution: The discussion on how the market for GNSS technology has evolved, including the impact of autonomous cars on the development and availability of multi-frequency GNSS chips, and how these advancements benefited broader applications beyond surveying and construction.
5. Precision vs. Accuracy in GNSS: An explanation of the difference between precision and accuracy in the context of GNSS technology, and the significance of each in various applications like drone mapping and volume measurements.
6. RTK and PPP Explained: A detailed explanation of Real-Time Kinematic (RTK) and Precise Point Positioning (PPP), including how they work, their applications, and their advantages and limitations.
7. The Future of GNSS Technology: Insights into the future directions of GNSS technology, including the challenges and potentials for achieving sub-centimeter accuracy with smartphones, the complementarity of GNSS and visual positioning systems, and the potential for GNSS technology to replace traditional surveying methods.
8. Global Utility of GNSS: A discussion on the importance of GNSS as a global utility, its indispensable role in modern technology and everyday life, and the potential consequences of GNSS failures.

More GNSS-related podcast episodes!

• From GNSS to VPS
• Reimagining GPS
• How Google Knows Your Location
• Past, Present and Future of GNSS
• SBAS – A base station in the sky

In Conversation

Building a GNSS Receiver in a Kitchen

Daniel: You’re the CEO and co-founder of Emlid, focused on high-precision GNSS receivers and software. Could you give us a brief introduction and your startup story?

Igor: We started the company about 10 years ago, when drones were becoming really popular and we were building drones for mapping while still at university. To make accurate maps, you need ground control points — known points on the ground — and to make those you had to use a high-precision or RTK GPS. That technology was very scientific and intimidating: a receiver could easily be $20,000, it required training and a separate computer running Windows software. We couldn’t stop thinking there had to be a more affordable way. Being a little naive, we decided to give it a go — in my kitchen we designed some boards, assembled them, rolled the software, and built our first GNSS receiver. It was nothing like the rugged field receivers with batteries and separate computers — just a board in some heat-shrink wrap. We had no way to scale it, so we launched a Kickstarter campaign — our first receiver, Reach, for just $200. People were used to $20,000 devices, and here were two guys working out of their kitchen offering the same technology for $200. We raised about $80,000, built our first batch, and our story starts from there.

Why High-Precision GNSS Used to Be So Expensive

Daniel: What was the secret sauce — how did you make it so much cheaper?

Igor: Honestly, it was possible to do it — the components were out there, and the open-source software was out there in open-source communities; no one had just tried to put it together as a product other people could use. Our main innovation was figuring out how to run it all on a really small device you could actually put on a drone or a robot. Previously you needed a computer to run the software for the calculations, plus extra hardware for the satellite reception, which made it very difficult to integrate. To get high accuracy reliably, you need a specialized chip that can receive multiple satellite systems and multiple signals on different frequencies from each satellite. Those specialized chips — like the ASICs people make for Bitcoin mining — cost millions of dollars in R&D up front. The companies that made that investment built equipment based on the chips just for themselves, and because the market was small — surveyors and some construction — prices stayed really high. The normal chip in your smartphone only receives one signal on one frequency from each satellite, and those cost about $10 because they’re produced in the millions.

How Autonomous Cars Changed the Market

Daniel: Did you develop a new chip, or did something change in the market?

Igor: For a change to happen, a new application had to appear where high-precision multi-frequency capability was a requirement and there was a massive volume of millions of units. The timing worked well for us — a few years after we started, the hottest topic was autonomous cars. Cars require higher accuracy than everyday applications because they want to know which lane they’re in, really reliably — that needed multi-frequency capability. Autonomous cars are a massive market, so suppliers who weren’t previously interested in this technology invested the money and developed chipsets that became available to everyone, much cheaper. It’s interesting how this technology evolved: it started with a specialized, expensive surveying use case, migrated to the much broader autonomous-car market — which made it far more affordable and even more robust — and what we’ve done is migrate those benefits of matured high-precision positioning back to surveyors, construction professionals, GIS, and drones.

Precision vs. Accuracy

Daniel: When we talk about survey-grade GNSS, we use “precision” and “accuracy” almost interchangeably. Is that a mistake?

Igor: You’re right that we tend to use both terms interchangeably, though if you dive into applications there are differences. With this technology you always have high precision, but you might not always get high accuracy. For drone mapping, if you want survey-grade results — boundaries — you absolutely want it to be really accurate, aligned not just to itself but to the broader reference system, so you know exactly where it is on a country level. But if you’re measuring volumes — a stockpile — you might not be worried about the pile being shifted one way or the other, as long as the volume is correct. So sometimes high precision is enough and high accuracy isn’t required. As for a magic cut-off — the devil is in the detail, depending on which metric you use. But in general, with high-precision RTK GPS you can expect to be within 1 centimetre, as long as your reference station isn’t too far away and you have a good sky view. We distinguish 1-centimetre-class devices, sub-metre-class devices, and your normal smartphone GPS, which is maybe within five metres. The sub-metre class is kind of dying — smartphones are getting closer to that precision, and centimetre accuracy is becoming more affordable, so if you can have centimetre accuracy for the same price and complexity, why not just do everything at that accuracy?

The Antenna Problem

Daniel: Beyond the chipset — what impact do the antenna, the software, and an IMU have?

Igor: The system has several critical components. First in the signal reception chain is the antenna. We want it to receive as much energy as possible for a really clean signal with minimal noise — and to receive multiple constellations and frequencies well, it needs to be about the size of a small plate. When you think about it, the antenna itself is bigger than the accuracy we’re trying to achieve, so we need to know exactly where within the antenna we’re receiving the signal. That virtual point where all the signal phases meet is called the phase center, and we want it to be very stable — on a millimetre scale — no matter the orientation of the device relative to the satellite, and the satellites are always moving. These antennas are quite specialized. Then there’s the receiver, producing raw data — ranges to the satellites and phase measurements — and the processing engine that makes the magic happen and outputs the position, plus batteries, Bluetooth, Wi-Fi, and the enclosure.

How RTK Works

Daniel: Could you give a high-level explanation of how RTK works?

Igor: A normal GPS in your smartphone reads signals from satellites — each satellite encodes a really precise timestamp of when the signal left, so when it arrives you can calculate the travel time, and with at least four satellites you derive your position, accurate to a metre or so. It’s only a metre because as signals travel through different layers of the atmosphere and ionosphere they get delayed differently, and a satellite on the left and one on the right send signals through completely different environments, accumulating errors. RTK uses two receivers — one stationary, the base station, and one moving, the rover. The satellites are about 20,000 kilometres away, while your two receivers are maybe 10 kilometres apart — so the signal paths are almost the same, the errors along the way are the same, and you can just subtract one from the other to remove the noise. RTK also uses integer ambiguity resolution: the signal is a wave about 20 centimetres long, and you can measure the phase — where you are within the wave — to millimetre accuracy, but you don’t know how many full wavelengths there are to the satellite. It’s like a mechanical clock: you know the time but not which day it is. A modern receiver tracks over 50 satellites at once, and an algorithm — which will certainly use a Kalman filter, among many estimated parameters — works out the integer wavelengths in an optimal way.

PPP, PPK, and Why Countries Drift

Daniel: And PPP and PPK?

Igor: PPK is the post-processing version of RTK — you don’t have a real-time connection between the two receivers; you record data on both and bring it together in specialized software afterwards. PPP is a bit different. Instead of assuming the errors are the same for two nearby receivers, what if we had a really good model of all the atmospheric layers around the Earth — like a cake wrapped around the planet, with different layers behaving differently — and could model each delay precisely? PPP builds a model for a very large region, ideally the whole globe, of exactly what those additional errors are for each signal at each moment in time. Once we get fully into a PPP future, the receiver will calculate a really precise position without relying on a second receiver nearby — it’s already doable in post-processing, though not yet in real time. RTK is still the king for positioning performance. And RTK has another advantage: because the reference stations are actual physical stations near you, if there’s a tectonic shift — if your country is slowly drifting, like Australia — your positioning drifts together with the tectonic plate, so your measurements stay in essentially the same place year over year. That matters for land boundaries: if you measured your land with a 1-centimetre Google Maps and came back 10 years later, you could be 30 centimetres off because the country has drifted. That’s why countries have their own frames of reference, with reference stations that drift along with the country.

The Future: Phones, Visual Positioning, and Surveying

Daniel: Could cell phones become a sub-centimetre positioning tool?

Igor: We might get there eventually — smartphone accuracy is definitely improving — but I don’t think we’ll see 1-centimetre survey-grade accuracy in smartphones within the next five years or so, for fundamental physical reasons. A good antenna is crucial for high-precision positioning, and there just isn’t space in a smartphone for one — and I don’t think manufacturers will see enough value to make it bigger and add a bigger battery for the constant calculations. It’s not just about receiving the signals — it’s the stability of the phase center. We rotate our phones however we please, and the smaller the antenna, the more difficult it is to keep that phase center stable. Even if you could, the phone is bigger than the accuracy you’re trying to measure, so you wouldn’t know which part to put on the point. For survey devices there’s a procedure called antenna calibration — you send your device to a special lab where a precise robotic arm rotates it into every orientation relative to each satellite, over many hours, to produce a calibration model — and people aren’t going to do that with cell phones.

Daniel: What about visual positioning — I saw a headline, “GNSS is dead, visual positioning is the new king.”

Igor: Saying something is dead is a sure way to get a lot of clicks. Visual positioning has a full 3D representation of the world stored in the device’s memory, and compares what you’re seeing to that model to understand where you are — a bit like GeoGuessr, but much faster and with better results. It sounds great for big cities, but it’s very challenging for remote areas with few features — in a cornfield, you look left, you see corn; you look right, you see corn. And a city changes less over time than vegetation. So the two technologies will complement each other really well: urban environments are challenging for GNSS because of reflections from buildings and obstructed sky view, so you augment GNSS with visual positioning there; in remote areas, at sea, or in the air, you rely on GNSS in those perfect conditions.

Daniel: Is there a world where we don’t need to send surveyors out to measure things?

Igor: For many jobs where you’d previously send a person to measure something, we’d no longer do that — creating a map could take weeks of walking and measuring; now you fly a drone and create it in hours. But for jobs where you need to locate something — to mark things up or install something — you’ll still use a high-precision positioning device to guide the person to the right place. GNSS is replacing the classic total-station way of measuring distances and angles for things like topographic surveys, because it’s so much faster — you get the coordinate right away, without first needing a known starting point. But if you need really high accuracy — in construction, where your error budget might be just a few millimetres — or you’re going into a mine or tunnel where GNSS doesn’t work, those classical instruments are here to stay, just becoming more of a niche tool for very specific jobs.