RTK (Real-Time Kinematic) positioning is a satellite-based technique that achieves centimeter-level accuracy by using correction data from reference stations to eliminate GNSS errors in real time. Where standard GNSS positioning delivers accuracy within several meters, RTK refines this to as little as one centimeter—a 100x improvement.
Historically called “RTK GPS” when GPS was the primary satellite constellation, modern RTK uses multi-constellation GNSS—including GPS, GLONASS, Galileo, and BeiDou—for superior satellite availability and positioning performance. Throughout this guide, we use “RTK” and “RTK positioning” to reflect current practice, though you may still encounter “RTK GPS” in legacy documentation.
RTK is foundational to precision industries including surveying, construction, agriculture, robotics, autonomous vehicles or any Physical AI application where centimeter-level accuracy makes the difference between success and failure. This guide covers how RTK works, what accuracy to expect, the types of RTK systems available, and how to get started.
For a detailed guide to how RTK correction data is generated and delivered, see our companion article: What Are RTK Corrections?
RTK One-Minute Explainer
RTK is | A GNSS positioning technique that uses real-time correction data from reference stations to achieve centimeter-level accuracy |
How it works | A base station at a precisely known location compares observed satellite signals against expected values, computes corrections, and transmits them to your rover (sometimes called a receiver) |
Accuracy | 1–3 cm horizontal and vertical, compared to 3–10 m with standard GNSS |
Convergence time | Seconds to achieve full precision (RTK Fix status)—far faster than alternatives like PPP |
Delivery | Corrections streamed via NTRIP over internet, or via L-Band satellite in areas without internet connectivity via cellular |
Key applications | Surveying, construction machine control, precision agriculture, autonomous vehicles, robotics, drone mapping, utility locating |
How RTK Works: The Technology Explained
RTK positioning relies on real-time communication between two components: a rover (your GNSS receiver in the field, often installed on a robot or vehicle) and one or more base stations (fixed reference stations at precisely surveyed locations).
The Basic RTK Process
- Base station observes satellite signals. Because the base station’s position is known to within one centimeter, it can calculate the errors affecting the satellite signals it receives. Errors include atmospheric delays, clock drift, orbital inaccuracies, and more.
- Corrections are computed and transmitted. The base station packages these error corrections into standard RTCM (Radio Technical Commission for Maritime Services) messages and transmits them to rovers in real time, typically via NTRIP over an internet connection.
- The rover applies corrections. Your receiver uses the correction data to cancel out the major error sources in its own satellite measurements, producing a position accurate to within a few centimeters.
Why RTK Is So Precise: Carrier Phase Measurements
The key to RTK’s precision lies in carrier phase measurements. Standard GNSS positioning relies primarily on code-phase (pseudorange) measurements—accurate, but limited to meter-level precision. RTK instead uses the carrier phase of the radio signal itself, which provides roughly 100x more measurement precision.
The challenge is that carrier phase measurements are ambiguous—the receiver can measure the fractional phase precisely, but doesn’t initially know the total number of complete carrier cycles between the satellite and the receiver. This is known as the integer ambiguity problem.
To understand why, consider how the math works. The GPS L1 carrier signal has a wavelength of approximately 19 cm. A receiver can measure where it sits within a single wavelength to sub-millimeter precision—but the signal looks identical every 19 cm. The true distance to the satellite equals some unknown whole number (N) of wavelengths plus the measurable fractional phase: distance = N × 19 cm + fractional phase. The receiver knows the fractional part precisely; the challenge is determining N, which may be in the millions. Getting N wrong by even one cycle introduces a 19 cm error—far too large for centimeter-level work.
RTK-capable receivers solve this using estimation algorithms that evaluate all possible integer combinations across multiple satellites and frequencies simultaneously. The most widely used approach, the LAMBDA (Least-squares AMBiguity Decorrelation Adjustment) method, transforms the ambiguity search space to make it computationally efficient, then validates the best candidate against alternatives using statistical ratio tests. When the receiver is confident in its solution—meaning one integer combination is statistically dominant—it declares Fix status and unlocks full centimeter precision. More satellites and more frequencies provide more independent equations, which is why multi-constellation, multi-frequency receivers resolve ambiguities faster and more reliably. For a deeper treatment of ambiguity resolution mathematics, see the GPS World technical library or Teunissen’s foundational work on the LAMBDA method.
Understanding GNSS Signal Errors
Error Source | Cause | Effect |
|---|---|---|
Signals slow down passing through the ionosphere and troposphere | Timing errors that translate to position errors | |
Satellite clock drift | Minor inconsistencies in onboard atomic clocks | Range calculation errors |
Orbital inaccuracies | Differences between predicted and actual satellite positions (ephemeris errors) | Geometry errors in positioning |
Signals reflecting off buildings, terrain, or surfaces before reaching the receiver | Interference and degraded accuracy |
Base stations account for all of these errors by comparing what they observe against what they should observe given their precisely known location. The resulting correction data allows rovers to eliminate these same errors from their own measurements. For a deeper dive, see RTK Corrections.
RTK Solution Statuses: Single, Float, and Fix
Status | What It Means | Typical Accuracy |
Corrections received, but integer ambiguities not yet resolved. | Sub-meter (decimeter-level) | |
All integer ambiguities resolved. Full RTK precision achieved. | 1–3 centimeters |
Fix status is the target for professional applications. Convergence time—how quickly a receiver transitions from Float to Fix—is typically just seconds with modern multi-constellation receivers and dense correction networks. This is a significant advantage over alternatives like PPP, which can take 20 minutes or more.
RTK Accuracy: What to Expect
Accuracy is RTK’s defining advantage. The numbers below represent what you can realistically achieve in practice—not theoretical maximums, but the performance that well-configured systems deliver in real-world conditions.
RTK Accuracy Specifications
Metric | RTK Positioning | Standard GNSS |
Horizontal accuracy | 1–3 cm (optimal conditions) | 3–10 m |
Vertical accuracy | 1–3 cm (optimal conditions) | 3–10 m |
Improvement factor | ~100x over standard GNSS | Baseline |
Factors That Affect RTK Accuracy
Factor | Impact |
Baseline distance | Accuracy degrades beyond ~35–50 km from a base station. Network RTK mitigates this through denser station coverage. |
Satellite geometry | More visible satellites from multiple constellations produce better PDOP and more reliable Fix. Multi-constellation receivers outperform GPS-only hardware. |
Signal environment | Open sky is best. Urban canyons, dense canopy, and terrain features block signals, reduce visible satellites, and delay time to Fix. |
Receiver quality | Multi-frequency, multi-constellation receivers resolve ambiguities faster and hold Fix more reliably. |
Atmospheric conditions | Ionospheric activity (especially solar events) and severe weather increase errors and slow convergence. |
Dense correction networks help mitigate both distance and environmental challenges by reducing baseline distances and providing redundant coverage.
Should you use a managed RTK service or deploy your own RTK Base Station?
Understanding RTK system architectures helps you select the right approach. The fundamental choice is between deploying your own base station and subscribing to a managed correction network.
Locally installed Base Station RTK
Aspect | Details |
How it works | One base station computes and transmits corrections based on its known position |
Effective range | Typically 30 km, with accuracy degrading at greater distances |
Best for | For use cases with most demanding precision and convergence requirements. Typically used in survey, damage prevention, drone inspection and delivery, construction, and mapping. |
Advantages | Full control, high accuracy within range, no subscription fees (after equipment and installation cost) |
Limitations | Limited coverage area, single point of failure, requires complex technical setup and ongoing maintenance |
Network RTK
Aspect | Details |
How it works | A network of base stations models errors across a region; corrections are interpolated for each rover’s location (often using VRS techniques) |
Coverage | Global depending on Network RTK service provider |
Best for | Multi-site operations, fleet deployments, applications requiring wide-area consistency and reliability |
Advantages | Fast set-up, consistent accuracy across large areas, redundancy, scalable, professionally maintained |
Limitations | Requires internet connectivity or L-band, ongoing subscription cost |
With VRS-based Network RTK, denser networks offer greater levels of precision due to proximity to nearest base station. This is why two networks advertising “centimeter accuracy” can perform very differently in practice—the one with tighter station spacing delivers more reliable results.
For most commercial and production applications, Network RTK offers the best combination of performance, reliability, and operational simplicity. Single base stations fit highly localized, fixed-site operations where deploying your own hardware makes economic sense. For more, see our NTRIP / RTK Service Provider Buyer’s Guide.
RTK vs. Other GNSS Correction Methods
RTK is one of several techniques for improving GNSS accuracy. Understanding the differences—including how corrections are represented and delivered—helps you choose the right approach for your application.
OSR vs. SSR: Two Ways to Represent Corrections
Every GNSS correction method delivers error corrections in one of two formats:
OSR (Observation Space Representation) delivers corrections as composite observations—all error sources lumped together—for the rover’s specific location. This is the traditional approach used by single-baseline RTK, Network RTK, and VRS services. OSR is straightforward, widely compatible with any RTK receiver, and places the computational burden on the correction provider rather than the rover. The trade-off: because corrections are location-specific, each user needs a unique data stream, which limits scalability.
SSR (State Space Representation) models each error source separately—satellite orbits, clocks, ionospheric delay, tropospheric delay—and sends those model parameters to the rover, which reconstructs its own correction. Because the same broadcast stream works for all users regardless of location, SSR scales efficiently to millions of devices. The trade-off: the rover needs more sophisticated processing firmware to interpret the individual error components.
A note on VRS and modeling: Network RTK services that use Virtual Reference Stations (VRS) do involve atmospheric modeling—the server interpolates ionospheric and tropospheric conditions across the network to synthesize observations for the rover’s location. But the model runs server-side; the rover still receives composite observations (OSR). This is different from SSR, where the model parameters themselves are sent to the rover.
For most users, this distinction matters when evaluating two things: receiver compatibility (OSR works with virtually any RTK receiver; SSR requires specialized firmware) and scalability (OSR requires a unique stream per user; SSR can broadcast one stream to all). For more on how corrections reach your receiver, see What is NTRIP?
Comparison of GNSS Correction Methods
DGNSS | Single-Baseline RTK | Network RTK (VRS) | Virtual Corrections (SSR-modeled, OSR-delivered) | PPP | PPP-RTK | |
Representation | OSR | OSR | OSR | SSR on backend → OSR to rover | SSR | SSR |
Correction type | Code-phase differential | Carrier-phase from nearest physical base | Carrier-phase from an emulated station or VRS (virtual reference station) | Standard RTCM (looks like single-baseline | Precise orbit + clock products | Orbit + clock + regional atmospheric models |
Uses atmospheric modeling? | No | No (relies on spatial correlation) | Yes (server-side interpolation) | Yes (full SSR model) | Orbits + clocks only (no regional atmosphere) | Yes (regional ionosphere + troposphere) |
Accuracy | 0.5–1 m | 1–3 cm | 1–3 cm | ~10 cm | 5–20 cm (post-convergence) | 2–10 cm |
Convergence | Instant | Immediate | Seconds | ~30 seconds | 15–30+ minutes | Seconds |
Receiver requirements | Any GNSS receiver | Dual-band RTK receiver | Dual-band RTK receiver | Dual-band RTK receiver | Specialized PPP firmware | Specialized PPP-RTK firmware |
Coverage | Near base station(s) | ~40 km from base | Within network footprint | Continental | Global | Regional–continental |
Best for | Asset tracking, fleet management, GIS | Survey, construction, precision ag, drones | Fleet operations within network coverage | Automotive, fleets at scale, IoT | Remote/offshore where convergence delay is acceptable | Automotive ADAS, mass-market autonomy |
Point One product | DGNSS | Network RTK |
|
Where Point One Fits
Point One Navigation offers two correction approaches that cover the most common precision positioning needs, both delivered through the Point One RTK Network—the world’s largest, professionally managed corrections network with thousands of base stations and 99.9% uptime.
True RTK is single-baseline RTK service, connecting your rover to the nearest physical base station in Point One’s network. It delivers 1–3 cm accuracy with immediate fix times—the gold standard for applications where maximum precision matters, like survey, construction, and drone operations.
Virtual RTK is another Point One service that uses SSR modeling internally—building atmospheric, orbit, and clock models the same way PPP-RTK services do—but converts the output back to standard RTCM messages that any RTK receiver can consume. This means you get the broadcast scalability and continental coverage of SSR without needing specialized receiver firmware. At ~10 cm accuracy with 30-second convergence, it’s designed for applications like automotive ADAS, fleet operations, and IoT where consistent wide-area coverage matters more than sub-centimeter precision.
For environments where GNSS signals are obstructed—urban canyons, tree canopy, tunnels—Point One’s Positioning Engine adds sensor fusion and dead reckoning to maintain precise location with or without open sky. And with the Point One GraphQL API, you can integrate provisioning, observability, and location data directly into your existing workflows.
Ready to test centimeter-level accuracy? Start a free 14-day trial—no credit card required. For enterprise or OEM volume pricing, contact sales.
For a detailed technical breakdown of all GNSS correction methods, see our GNSS Correction Methods Compared guide.
RTK Applications Across Industries
RTK positioning enables capabilities impossible with standard GNSS accuracy. Across industries, the common benefits are reduced waste, optimized resource use, improved safety, and increased automation. Some common industries that benefit from RTK are:
Precision Agriculture
RTK enables automated tractor guidance with centimeter precision, eliminating overlaps and gaps in planting, spraying, and harvesting. Variable rate application systems use precise positioning to adjust inputs based on exact field location, increasing yields while reducing costs. Field mapping at centimeter resolution helps farmers track crop performance zones and optimize planting patterns over time. See how TRIC Robotics uses RTK for autonomous pest control on strawberry farms.
Construction and Machine Control
Machine control systems for excavators, graders, and bulldozers use RTK to follow design specifications in real time—operators see their position relative to design plans without grade stakes or manual measurements, increasing productivity and reducing costly rework. RTK also enables accurate site layout, volumetric earthwork calculations, and as-built documentation for quality control. Benchmark Tool & Supply integrates RTK corrections for centimeter-level construction accuracy.
Surveying and Mapping
RTK provides survey-grade accuracy for topographic, boundary, and GIS surveys without post-processing delays. Surveyors verify accuracy in the field and recollect questionable points in a single visit—completing in one efficient session what once required multiple trips. For example, solutions that use RTK to prevent utility damage resulting from construction. Seamless integration with CAD and GIS software means field data flows directly into design and analysis workflows.
Autonomous Systems and Robotics
Self-driving vehicles, agricultural robots, delivery drones, and warehouse automation all rely on RTK for safe, precise navigation. In dynamic environments, centimeter-level accuracy is essential for obstacle avoidance, path following, and split-second decisions where even small positioning errors can mean the difference between safe operation and collision. RTK is often combined with IMU and other sensors in an INS approach, where dead reckoning bridges gaps when GNSS signals are temporarily unavailable. See how RTK is used for autonomous ground support at airports or how RTK powers autonomous landscaping and mowing. For developers, see Absolute vs. Relative Navigation.
RTK Equipment: Receivers and Hardware
Not all GNSS receivers support RTK. Selecting the right hardware involves evaluating several key specifications that determine real-world performance.
Specification | What to Look For | Why It Matters |
Constellation support | More satellites → better geometry → faster, more reliable Fix | |
Frequency bands | Multiple independent measurements per satellite → faster ambiguity resolution | |
Update rate | Generally 1–10 Hz depending on application | Higher rates essential for drones and autonomous vehicles; lower rates fine for surveying |
Convergence time | Under 5 seconds for professional equipment | Minimizes downtime on power-on or signal recovery |
Connectivity | NTRIP over cellular, radio link, API access | Determines how receiver accesses correction data and integrates into workflows |
When using a network RTK service, setup is straightforward: connect to the provider’s NTRIP mountpoint with any compatible receiver and begin receiving corrections.
Setting Up RTK: Your Options
There are three main paths to accessing RTK corrections, each with distinct trade-offs in cost, complexity, and coverage. When evaluating RTK providers, keep in mind that “NTRIP service provider” and “RTK correction provider” are effectively the same thing—NTRIP is the standard protocol used to deliver RTK corrections over the internet.
Approach | Advantages | Limitations | Best For |
RTK Service Provider subscription | Broad coverage, no hardware to deploy, fast setup, professional support, high uptime, scalable | Ongoing subscription cost, requires internet unless using L-Band | Most commercial applications—robotics, agriculture, construction, multi-site operations |
Own base station | One-time equipment cost, full control, no subscription | Significant upfront cost ($2K–$15K+), technical expertise required, 10–50 km range, single point of failure | Dedicated fixed-site operations with in-house GNSS expertise |
Public CORS networks | Free or minimal cost (see directory here) | Variable coverage, no SLAs, limited support, inconsistent performance and uptime | Traditional surveying, occasional field work, experimentation |
Before committing to a DIY base station, see Is Building Your Own RTK Worth It? For guidance on selecting a network provider, see our NTRIP / RTK Service Provider Buyer’s Guide.
Optimizing RTK Performance
Achieving the best RTK results in the field requires attention to practical considerations. Most performance issues stem from environmental factors and configuration rather than equipment limitations.
Best Practice | Details |
Maintain clear sky visibility | Plan operations to minimize time in heavily obstructed areas—near tall buildings, under dense canopy, or in deep terrain cuts. |
Minimize baseline distance | Stay within 10–50 km for single base setups. Network RTK maintains accuracy over larger areas but shorter baselines still perform best. |
Avoid electromagnetic interference | Position antennas away from power lines, generators, motors, and radio transmitters. |
Optimize antenna placement | Mount with clear sky view, away from metal surfaces causing multipath errors. Keep level and secure. |
Configure equipment correctly | Verify update rates, coordinate reference system, datums/epochs, and correction data sources. Many issues stem from configuration, not equipment. |
RTK vs. Standard GNSS
For readers evaluating whether RTK is necessary for their application, here’s how it compares to the standard GNSS built into consumer devices like smartphones and vehicle navigation systems.
Feature | Standard GNSS | RTK Positioning |
Accuracy | 3–10 meters | 1–3 centimeters |
Cost | Free (generally built into devices) | Subscription or equipment investment |
Infrastructure | None required | Base station or network access |
Convergence | Instant | Seconds (RTK Fix) |
Use cases | Navigation, general location, recreation | Surveying, construction, autonomy, precision agriculture |
Special hardware | Standard GNSS chip | RTK-capable GNSS receiver |
Standard GNSS is adequate for navigation and general location services. RTK is necessary when accuracy directly impacts safety (autonomous vehicles), efficiency (machine control), compliance (surveying), or product quality (precision agriculture). The investment typically pays for itself when a single positioning error prevented exceeds the cost of the service.
RTK FAQs
How accurate is RTK?
RTK typically provides 1–3 cm horizontal and vertical accuracy under ideal conditions. Accuracy depends on baseline distance, signal environment, and receiver specification. Under optimal open-sky conditions with a nearby base station, 1 cm horizontal accuracy is achievable.
How long does RTK take to converge?
With modern multi-constellation receivers and a dense correction network, RTK can achieve Fix status in seconds. In challenging environments (partial sky obstructions, high ionospheric activity), convergence may take up to 30 seconds or longer.
How much does RTK cost?
RTK pricing varies by service provider and approach. Network RTK subscriptions typically range from $50–$200+ per device per month. Building your own base station involves $2,000–$15,000+ in upfront hardware plus ongoing maintenance. Public CORS networks are often free but lack reliability guarantees. RTK-capable receivers range from a few hundred dollars for basic modules to several thousand for professional multi-frequency units.
What’s the difference between RTK and DGPS?
DGPS (Differential GPS) improves standard GPS accuracy to roughly 0.3–1 meter using code-phase corrections. RTK, on the other hand, uses carrier-phase measurements to achieve 1–3 cm—an order of magnitude better. RTK also works with multi-constellation GNSS, providing better satellite availability and reliability.
What’s the difference between RTK and PPK?
Both RTK and PPK (Post-Processing Kinematic) use carrier-phase measurements for centimeter accuracy, but differ in when corrections are applied. RTK applies corrections in real time; PPK records raw observations and corrects afterward. RTK is essential for real-time applications (autonomous navigation, machine control). PPK suits drone mapping and aerial surveying where data is corrected after landing. Some workflows combine both for real-time guidance plus post-processed refinement.
Can RTK work without internet?
Yes. Single base station RTK can deliver corrections via radio link (UHF/VHF), though range is limited. For wider coverage, L-Band satellite delivery broadcasts corrections from geostationary satellites directly to the rover—ideal for agriculture, mining, and remote operations without cellular coverage. Network RTK via NTRIP does require an internet connection.
How far can RTK work from a base station?
Performance typically declines beyond 35–50 km, with longer baselines causing slower convergence and reduced Fix reliability. Base station density is important in any RTK network, whether using single-baseline or Network RTK which models a VRS (Virtual Reference Station) from multiple physical base stations to maintain accuracy across wide areas.
Do I need a special GNSS receiver for RTK?
Yes. RTK requires a receiver capable of carrier-phase measurements and integer ambiguity resolution. Consumer GNSS chips (smartphones) do not support RTK. RTK receivers range from affordable single-frequency modules to professional multi-frequency units, and must support NTRIP or another correction input method.
Can RTK work indoors or under heavy tree cover?
RTK requires line-of-sight to GNSS satellites. It generally does not work indoors. Under heavy canopy, performance degrades—Fix may not be achievable. Multi-constellation receivers perform better in partially obstructed environments. For indoor or heavily obstructed applications, RTK is often combined with IMU, lidar, or cameras in a sensor fusion approach.
Why use GNSS instead of just GPS for RTK?
GPS is one constellation of four. GNSS includes GPS, GLONASS, Galileo, and BeiDou—giving modern receivers access to 100+ satellites instead of ~30. This improves geometry, speeds convergence, and improves reliability in challenging environments. “RTK GPS” is a legacy term; modern RTK is multi-constellation by default.
Get Started with Point One RTK
Point One RTK Network is a commercial RTK correction network built for production applications—from robotics and autonomy to precision agriculture and construction.
- 3,000+ base stations across the US, EU, UK, Canada, and AUS & NZ
- 99.9% uptime SLA with real-time network monitoring
- 1–3 cm accuracy with convergence in seconds
- Single mountpoint for global operations—no manual station selection
- Multi-constellation GNSS support (GPS, GLONASS, Galileo, BeiDou)
- Hardware-agnostic compatibility with major GNSS receivers
- Developer-first GraphQL API, SDKs, and documentation
Setup takes five minutes: connect to the Point One NTRIP mountpoint with any compatible receiver and start receiving centimeter-accurate corrections.
Start your free trial → for a two-week free trial of Point One’s RTK corrections service
Contact sales to discuss enterprise deployments, coverage needs, or custom integrations.