Inertial Magnetic SLAM Systems Using Low-Cost Sensors

Spatially inhomogeneous magnetic fields offer a valuable, non-visual information source for positioning. Among systems leveraging this, magnetic field-based simultaneous localization and mapping (SLAM) systems are particularly attractive because they can provide positioning information and build a magnetic field map on the fly. Moreover, they have bounded error within mapped regions. However, state-of-the-art methods typically require low-drift odometry data provided by visual odometry or a wheel encoder, etc. This is because these systems need to minimize/reduce positioning errors while exploring, which happens when they are in unmapped regions. To address these limitations, this work proposes a loosely coupled and a tightly coupled inertial magnetic SLAM (IM-SLAM) system. The proposed systems use commonly available low-cost sensors: an inertial measurement unit (IMU), a magnetometer array, and a barometer. The use of non-visual data provides a significant advantage over visual-based systems, making it robust to low-visibility conditions. Both systems employ state-space representations, and magnetic field models on different scales. The difference lies in how they use a local and global magnetic field model. The loosely coupled system uses these models separately in two state-space models, while the tightly coupled system integrates them into one state-space model. Experiment results show that the tightly coupled IM-SLAM system achieves lower positioning errors than the loosely coupled system in most scenarios, with typical errors on the order of meters per 100 meters traveled. These results demonstrate the feasiblity of developing a full 3D IM-SLAM systems using low-cost sensors and the potential of applying these systems in emergency response scenarios such as mine/fire rescue.

💡 Research Summary

This paper presents a novel approach to Simultaneous Localization and Mapping (SLAM) that operates without visual or wheel-odometry data, instead leveraging the spatial variations of ambient magnetic fields. The proposed Inertial Magnetic SLAM (IM-SLAM) systems utilize only low-cost, ubiquitous sensors: an Inertial Measurement Unit (IMU), an array of magnetometers, and a barometer. This sensor suite makes the system inherently robust in low-visibility or featureless environments where traditional visual or LiDAR SLAM fails.

The core innovation lies in the strategic use of magnetic field modeling at two distinct scales. First, a local magnetic field model, based on a first-degree polynomial, captures the magnetic field within the immediate area covered by the magnetometer array at each timestep. This model is used within a Magnetic field-Aided Inertial Navigation System (MAINS) to correct velocity drift from the IMU, effectively substituting for the external low-drift odometry required by prior magnetic SLAM methods. Second, a global magnetic field model, employing a computationally efficient reduced-rank Gaussian Process (GP) approximation with sinusoidal basis functions, builds a map of the magnetic field across the entire explored environment.

The authors propose and compare two architectural variants for integrating these components. The Loosely Coupled IM-SLAM system runs the MAINS (using the local model) and a global magnetic field SLAM estimator as two separate processes, fusing their outputs. In contrast, the Tightly Coupled IM-SLAM system integrates the IMU state, the local magnetic field model parameters, and the global magnetic field model parameters into a single, high-dimensional state vector. This unified state is estimated within a single Bayesian filter (like an Extended Kalman Filter), allowing all sensor measurements and model uncertainties to be processed jointly in a theoretically optimal framework.

Extensive experiments were conducted in real-world, multi-floor indoor environments. The results demonstrate that the Tightly Coupled system consistently outperforms the Loosely Coupled one, achieving end-to-end positioning errors on the order of a few meters per 100 meters traveled. The barometer is shown to be crucial for stabilizing altitude estimation in 3D spaces. The work conclusively proves the feasibility of building a full 3D SLAM system using only low-cost inertial and magnetic sensors, eliminating the dependency on cameras or wheel encoders. This breakthrough paves the way for reliable navigation systems in challenging GNSS-denied scenarios such as emergency response, mine exploration, or firefighting, where visibility is poor. The paper contributes to reproducible research by making both the datasets and algorithm implementations publicly available.