A Pedestrian SLAM Scheme Grounded in Inertial-Based Map and Magnetic Field Map

Inertial pedestrian navigation systems (IPNS) face challenges in practical applications due to inherent noise and bias instability of inertial sensor. To construct a high-precision, prior information-independent, low computational complexity inertial-based pedestrian navigation system, this article proposes the inertial magnetic field hybrid map simultaneous localization and mapping (IMHM-SLAM) scheme. First, to improve the system’s precision, IMHM-SLAM scheme utilizes inertial-based maps to represent the passable areas within enclosed environments and correct most localization errors caused by inertial sensors. Additionally, magnetic field maps are employed to describe the environmental characteristics of the passable areas, thereby resolving the ambiguity associated with the association between different regions. Second, to enable the system to operate independently of prior information (referred to as pedestrian activity ranges in this study), a scalable magnetic field map model is introduced. This model is based on recursive Gaussian process regression (RGPR) and map basis vector update methods. Third, in order to reduce the computational complexity of magnetic field map updates when pedestrians move in larger areas, a local update strategy is proposed. This strategy ensures a fixed upper limit on computational complexity. Finally, simulations and field trial evaluations demonstrate that the IMHM-SLAM scheme, utilizing the low-cost inertial sensor and magnetometer in smartphones, achieves an average positioning accuracy of 1.41 m in a small-area (500
$\text{m}^{{2}}$
) field trials and 2.56 m in a large-area (2600
$\text{m}^{{2}}$
) field trials.