RTK GPS Demystified: What It Takes to Reach Centimeters

RTK (Real-Time Kinematic) GPS is a satellite-positioning method that uses an external correction stream to cancel positioning errors and achieve centimeter-level accuracy in real time. The correction can come from a local base station, a private setup, or a public network service.

Consider two people walking across a campus with ordinary phones. A phone position is often only accurate to about 5 to 10 meters. Both phones see roughly the same satellites through roughly the same sky, so they tend to be wrong in a similar direction. Now put one phone on a tree whose position is already known. The phone at the tree can compare its known location with what the satellites currently say, work out how much the satellite position is off right now, and send that correction to the moving phone nearby.

That picture is simplified. Strictly speaking, it describes position correction more than the full RTK mechanism. But the intuition is the same: one known reference helps the moving device subtract a known error. RTK uses the same idea with a finer ruler — the carrier phase of the satellite signal — to push the result from meters toward centimeters.

For a person looking at a map, a few meters may only mean walking a few more steps. For a machine that must act on a coordinate, a few meters can be the difference between doing the job and missing it completely.

Standard GPS vs RTK: Why Ordinary GPS Stops at Meters

Before evaluating if RTK makes sense for your project, you first need to pinpoint exactly what standard GPS is missing. The short version is simple: standard GPS estimates where you are; RTK estimates where you are and then corrects that estimate using an outside reference.

What standard GPS gets you

Standard GPS, or more broadly standard GNSS, calculates position from satellite signals. Your receiver measures how long signals take to arrive, then uses those measurements to estimate your location. That is enough for driving directions, phone maps, asset tracking, and many outdoor apps, but the result is usually in the meter-level range.

The reason is not that the satellites are careless. The signal path is affected by clock errors, atmosphere, satellite geometry, reflections, and receiver-side limitations. For a deeper explanation of the basic satellite-positioning model, start with our guide on how GNSS works. This page only needs the conclusion that standard positioning leaves you with meters of uncertainty.

What RTK adds

RTK adds an external correction stream. Instead of trusting the raw satellite solution alone, the moving receiver applies corrections that describe the errors seen from a known reference. The result is not just “GPS with a better chip.” It is a different workflow: satellite signals plus corrections plus a receiver that can use them.

The practical difference is easy to picture. Meter-level positioning may tell a robot which lane of a field it is in. Centimeter-level positioning can tell it whether it is still on the same row, boundary, pass line, or sampling point.

Free signal, paid correction

One cost detail is easy to miss: the satellite signal is free; the correction that turns it into centimeters usually is not. Anyone can receive GNSS signals. RTK adds a correction source, and that source often depends on a provider, an account, a local base station, or a public correction network.

Some regions have public or free correction sources, and some projects run their own base. Many real deployments still involve a paid correction service or a maintained local setup. That means upgrading from standard GPS to RTK is not only a receiver decision. It also adds an ongoing dependency on correction availability.

How RTK Works: A Reference Receiver and a Stream of Corrections

A rough mental model of this workflow is the best way to predict when RTK will hold a centimeter fix and when it will struggle. RTK is easiest to understand as a relationship between a known receiver, a moving receiver, the same satellites, and a correction stream.

RTK Correction Loop

01

Reference receiver

A receiver sits on a known coordinate and compares that known position with the current GNSS solution.

RTCM / NTRIP

Correction stream

The shared error is sent to the moving receiver in real time.

02

Rover receiver

The rover applies the correction and reports a high-precision position when the RTK solution is ready.

Fixed Centimeter-grade state under good conditions.

Float Still resolving; not final yet.

In formal RTK vocabulary, the receiver on the known point is the base or base station; in the campus example, it is the phone fixed to the tree. The moving receiver is the rover; it is the phone walking nearby. The diagram above is still a simplified model. True RTK also relies on carrier-phase measurements and ambiguity resolution to reach the centimeter level, but those mechanics belong in a deeper engineering guide. For a beginner overview, the core intuition is enough: a reference receiver helps the moving receiver subtract shared errors in real time.

How the correction moves from base or network to rover is a separate topic. In many internet-based deployments, it moves through NTRIP; for details on that transport layer, read our guide to NTRIP and RTK correction streaming. The correction messages themselves are usually RTCM, which is covered in RTCM Unpacked. For this page, the important idea is only that the rover needs a usable correction stream.

Fixed vs Float: What You'll Actually See

Your RTK receiver will normally report a solution state. Fixed means the receiver has resolved the carrier-phase solution and is confident enough for centimeter-grade work under suitable conditions. Float means the solution is still being resolved; in practical field experience, it is often closer to sub-meter or decimeter-level positioning than true centimeter positioning, depending on sky view, corrections, baseline, antenna setup, and multipath.

If your rover says Float, do not rush the machine into precise action just because numbers are arriving. Give the receiver time to converge, check the antenna environment and correction age, and treat Float as a useful warning: the receiver is working, but it is not ready for final centimeter decisions.

How Accurate Is RTK GPS, Really?

The phrase “centimeter accuracy” is useful only when you know what kind of promise it is. In good conditions, RTK can support tasks that meter-level GPS cannot: returning to the same sampling point, following a field row, checking a boundary, or aligning a machine pass to a planned path.

The basic accuracy ladder is easiest to compare in one view:

The headline centimeter is a ceiling, not a guarantee. Real accuracy moves with multipath, blockage, distance from the reference, correction quality, receiver algorithms, and antenna setup. A receiver can also output a clean-looking coordinate without giving your application a full uncertainty picture.

A spec such as 1 cm + 1 ppm means there is a fixed centimeter term plus a distance term. The ppm part adds about 1 millimeter for every kilometer between the rover and the reference, so long baselines are less forgiving than short ones.

The practical rule is not to fear RTK. It is to treat it like an engineering signal, not a magic label. If a task is safety-critical, expensive to redo, or tied to fixed assets, verify key points and understand the accuracy indicators your receiver exposes. For the statistics behind accuracy terms such as CEP, RMS, and confidence, see GNSS Accuracy Decoded.

When Is RTK Worth It — and When Is It Not?

RTK is not automatically better for every task; it is better when your task can actually use the extra precision. If the work only needs to know which area something is in, meter-level positioning may be enough. If the work needs repeatable lines, boundaries, or points, RTK starts to matter.

Use this as the first decision filter before choosing hardware, corrections, or software:

There is also an important distinction between positioning and navigation. Positioning answers “where am I right now?” Navigation answers “how do I move along this planned path and perform the work?” A centimeter-grade point is not the same as a complete guidance system.

That is why RTK is especially valuable for path-following tasks: a drone flight line, a tractor pass, a robot route, a sampling grid, or a construction layout point. The requirement is not simply to know the current position. The requirement is to repeat, align, record, or act on that position in a way the rest of the system understands.

A centimeter coordinate only becomes valuable when the downstream application can consume it. You still need the right coordinate frame and datum, a basemap or GIS layer, a planned path, and a workflow that records or uses the coordinate correctly. A perfect fix in WGS84 can still look wrong on a local map if the coordinate framework is mismatched; that topic belongs in Beyond WGS84.

This guide is vendor-neutral: the same reasoning holds for any RTK receiver. RTK gives you position. It does not replace antenna placement, correction monitoring, coordinate management, mapping, planning, or degradation logic in the application.

What RTK Is Used For

RTK becomes easier to judge when you compare it with a real use case near your own. The common thread is not “more accurate GPS” in the abstract. It is repeatable outdoor work tied to real coordinates.

Agriculture. RTK is used for field boundaries, plot revisit, soil sampling, guidance lines, controlled traffic, implement positioning, and repeated seasonal measurements. The value is repeatability: the ability to come back to the same row, line, or point instead of only the same general area. For agricultural repeatability and revisit workflows, see RTK GPS for Field Trials.

Robotics and autonomous machines. Outdoor robots use RTK as a global absolute-position layer. It helps anchor routes, boundaries, work zones, and fleet-level maps, but it normally works with other sensors rather than replacing them. For a robot-specific view, see RTK GPS for Robotics.

Surveying, mapping, and field data collection. RTK helps teams collect points that must line up with assets, parcels, utilities, ground control, or GIS layers. The receiver output matters, but so do the coordinate frame, the field app, and the project standard.

Construction and earthmoving. RTK supports layout, machine guidance, grading reference, stockpile measurement, and site progress checks. The accuracy is useful when moving from “near this area” to “follow this line or place this point.”

What You Need to Run RTK

RTK is a small system, not a single setting. At minimum, you need an RTK-capable receiver and a correction source. The receiver must be able to process carrier-phase measurements and apply corrections; the correction source must be reachable, compatible, and reliable enough for the job.

The receiver can be a module inside a product, an integrated receiver, a survey rover, or a compact external unit connected to another device. If you are deciding between a board-level module and a packaged receiver, see RTK Module vs RTK Receiver. This page only needs the basic rule: the receiver has to be designed for high-precision GNSS, not just ordinary navigation.

For machine projects, the receiver also has to survive the installation: antenna placement, enclosure, cable strain, power stability, and repeatable mounting can matter as much as the chipset inside.

The correction source may come from a network service, a public correction network, a local base station, a radio link, or an internet/NTRIP workflow. How to choose and operate that source is a separate article. Here, the key point is that the rover needs corrections continuously enough to maintain the solution state your application depends on.

Can an ordinary device just do high precision?

Usually not. High precision is designed into the hardware from the start: antenna quality, RF path, multi-frequency capability, carrier-phase measurement, firmware behavior, and correction handling all matter. A regular phone or navigation terminal cannot normally become a centimeter RTK rover just by installing an app, buying a service, or receiving a software update.

Centimeter accuracy is designed into the hardware, not added by an app or a subscription. Software and services matter, but only after the device is physically capable of producing and using the measurements RTK needs.

Conclusion

RTK GPS is best understood as standard satellite positioning plus real-time correction. A known reference helps remove errors the rover also sees, and carrier-phase measurement gives the rover a finer ruler. That combination is what moves positioning from meters toward centimeters.

Once those boundaries are clear, RTK becomes easier to evaluate. It is not a magic accuracy label; it is a practical way to make satellite positioning useful for repeatable outdoor work.

Frequently Asked Questions