Sensor Fusion for Localization
Radio, inertial sensing, geomagnetism, vision and GNSS work together to produce a position and orientation estimate suited to real conditions, even as the quality of individual sources changes.
Localization changes with the environment
NIST research on RSS shows why indoor environments are difficult: walls, people and reflective surfaces alter radio propagation, while a signal map can become sensitive to changes in the site. A GiPStech technical article on Bluetooth beacons illustrates in practical terms why RSSI alone is not a stable distance reference; it is a company source, not independent validation.
Other sources also have favorable conditions and limits: inertial estimates accumulate drift over time, vision depends on the scene and lighting, geomagnetism requires site characterization, and GNSS quality declines when the sky is obstructed. A design must therefore understand its context instead of assuming a universal source.
Sources contribute according to their availability and quality in the operating context.
One coherent estimate from complementary observations
An IEEE survey of fusion-based indoor systems describes the value of combining techniques and sources when complex electromagnetic environments reduce the accuracy and robustness of individual approaches.
Within the GiPStech core, sources are not merely placed side by side. Available measurements feed a common estimate: absolute references help constrain drift, motion observations maintain the trajectory between corrections, and current signal quality determines which contributions remain usable.
Each source answers a different question
A scientific review of magnetic-field positioning documents both the value of indoor magnetic anomalies and the challenges of calibration, interference and map construction. The same discipline applies to every source: describe its contribution together with its operating conditions.
Radio frequency
Provides references tied to beacons, networks or tags. Coverage, multipath and obstruction depend on the environment.
IMU and inertial
Estimates motion, steps and orientation changes with local continuity, but needs corrections to constrain drift.
Geomagnetism
Matches a building's magnetic variations with a reference map without introducing dedicated transmitters.
Vision
Recognizes scene geometry and features to estimate pose and motion when the camera, lighting and context permit.
GNSS
Offers a global outdoor reference; buildings, roofs and urban canyons can reduce availability and quality.
The system changes its balance without losing context
Research on GNSS, visual and inertial fusion shows how complementary measurements can support a smooth local estimate and a consistent global reference through transitions and signal degradation. These scientific results describe a general principle, not a measurement of GiPStech performance.
In the sensor fusion engine, a temporarily weak source can receive less weight while other sources maintain the trajectory or provide a new reference. When the source becomes useful again, the system can reintroduce it into the estimate. Actual behavior depends on configuration, map quality and project requirements: designed continuity does not mean guaranteed continuity under every condition.
Configuration starts with requirements
Modularity makes it possible to choose which sources to activate and how to combine them. Required accuracy, expected continuity, available infrastructure, devices, power use and site scale must be assessed together; no configuration is universally best.
Define the outcome
Clarify targets, areas, update needs, orientation and tolerance for interruptions.
Assess the environment
Assess available radio, geometry, lighting, materials, sky visibility and expected changes.
Compose the modules
Select sources that provide complementary references and define fallback behavior.
Integrate and verify
Connect the engine to the application or platform and measure behavior in the real scenario.
A technology core for different targets and integrations
The Technology page describes the wider GiPStech framework: the core can support the localization of people and devices, vehicles and mobile systems, assets and spaces by selecting the modules suited to the project.
GiPStech Solutions show how position data supports applications for airports, industry, healthcare, retail and public spaces. Asset2Work is one possible downstream operational platform; sensor fusion remains the technology subject of this page.
The work with HERE Positioning documents the integration of GiPStech inertial and sensor fusion technology into a third-party ecosystem. Contact GiPStech to discuss current integration options and platform compatibility for your project.
Explore related technologies: multi-sensor indoor positioning, adaptive hybrid localization, and Visual Positioning and VIO.
Tell us about the environment and what you need to locate
Share the targets, areas, available infrastructure, and continuity and integration requirements. We can help define a configuration that can be verified in the field.
