Introduction
GNSS is amazing, until you drive into a tunnel, walk through a dense forest, or fly behind a building. Then you lose the signal. But what if your device could "remember" where it was going and keep navigating anyway? That's what INS does.
What Is INS?
INS stands for Inertial Navigation System. It uses accelerometers (which measure acceleration) and gyroscopes (which measure rotation). By measuring how you accelerate and turn, INS can calculate your position relative to a known starting point, no external signals needed.
The Problem: INS Drifts
INS has a fatal flaw: drift. Accelerometers have tiny errors. Integrate acceleration to get velocity and errors grow. Integrate velocity to get position and errors grow even faster.
The Solution: GNSS + INS
GNSS and INS are perfect complements, each covers the other's weakness.
| System | Strengths | Weaknesses |
|---|---|---|
| GNSS | Stable long-term, absolute positioning | Intermittent, can be blocked |
| INS | Continuous, short-term accurate | Drifts over time |
Together: GNSS calibrates INS (corrects drift), and INS fills GNSS gaps (keeps working when signals are lost).
How Integration Works
- Loosely coupled: GNSS provides position; INS provides position changes. Simple, works well, but if GNSS is lost, INS drifts until GNSS returns.
- Tightly coupled: Raw GNSS measurements (pseudorange, carrier phase) fed directly into the INS filter. Can use even 1 satellite to help INS. Better performance in challenging environments.
- Deeply coupled: INS helps GNSS tracking loops maintain lock in high dynamics. Used in military and aerospace applications.
What Happens During Outages
Scenario: driving through a tunnel.
- Second 0: GNSS good, INS calibrated
- Seconds 1–10: GNSS lost; INS navigating on its own. Position error growing slowly. Quality indicator shows "coasting".
- Second 11: Exit tunnel, GNSS returns. Filter compares INS position with GNSS, updates INS error model, and recalibrates for the next outage.
Integration Levels by Application
| Application | INS Quality | Integration | Outage Tolerance |
|---|---|---|---|
| Smartphone | MEMS (poor) | Loose | 5–10 seconds |
| Car navigation | MEMS + wheel sensors | Tight | 30–60 seconds |
| Drone | Tactical | Tight | 1–2 minutes |
| Surveying | Tactical | Tight (with PPK) | Minutes |
| Aviation | Navigation-grade | Deep | Hours |
| Missile | Navigation-grade | Deep | Hours |
Real-World Examples
- Autonomous vehicles: Need continuous positioning; INS fills urban canyon gaps; wheel sensors help too
- Drone mapping: INS bridges moments when GNSS is lost, essential for consistent image geotagging
- Underground mining: No GNSS at all; INS with periodic updates from surveyed points
- Pedestrian navigation: Smartphone INS helps when walking through buildings
Sensor Fusion
Modern systems use more than just GNSS+INS. A Kalman filter combines magnetometer (compass), barometer, wheel odometry, and camera/LiDAR with GNSS/INS for the optimal position estimate.
Vital Points
- INS provides continuous navigation during GNSS outages
- GNSS calibrates INS to prevent drift
- Integration levels range from loose to deeply coupled
- Outage tolerance depends on INS quality, seconds to hours
- Essential for autonomous vehicles, drones, and any reliability-critical application
- Sensor fusion adds more sources for even better performance