Midterm colloquium Gustaaf Upperman

Title: One-Dimensional Magnetic SLAM with Loop Closures for Indoor Localization Using PDR with Non-Foot-Mounted IMUs

Supervisor: Dr. Manon Kok

Abstract: Accurate indoor localization is essential for applications such as emergency response, healthcare, and navigation in large indoor spaces. Traditional methods like Global Navigation Satellite System (GNSS) are ineffective in indoor environments due to signal obstructions, necessitating alternative approaches. Pedestrian Dead Reckoning (PDR) is a commonly used indoor localization technique that utilizes the Inertial Measurement Unit (IMU) to estimate position and orientation. However, IMU-based localization suffers from integration drift, leading to accumulated position errors. Foot-mounted IMUs benefit from Zero Velocity Update (ZUPT) to mitigate drift, but ZUPT is unavailable for non-foot-mounted IMUs. Despite this limitation, non-foot-mounted IMUs are more practical for real-world applications, as they can be integrated into wearable devices, smartphones, and other personal accessories without affecting user movement. Additional localization techniques must be explored to compensate for the lack of ZUPT. One promising approach is Simultaneous Localization and Mapping (SLAM), which incorporates additional spatial information to correct trajectory drift. This literature review investigates how one-dimensional magnetic SLAM with loop closures can enhance PDR accuracy using non-foot-mounted IMUs. This approach is lightweight, unlike traditional SLAM techniques that require extensive mapping. One-dimensional magnetic SLAM leverages the natural variations in indoor magnetic fields caused by structural ferromagnetic materials as distinctive features for localization. These magnetic signatures are utilized for loop closure detection, helping to correct accumulated drift in PDR. The first part of this review examines PDR systems that use non-foot-mounted IMUs, focusing on step detection, step length estimation, and heading determination using quaternion-based filtering. The second part explores one-dimensional magnetic SLAM, highlighting different methods for detecting and validating magnetic loop closures. Finally, sensor fusion methods like the Extended Kalman Filter (EKF) and Pose Graph Optimization (PGO) are reviewed for integrating PDR with loop closure information. This literature survey concludes that one-dimensional magnetic SLAM with loop closures enhances map-building when a dedicated surveyor generates frequent loop closures. However, further research is needed to assess its effectiveness for indoor localization without a dedicated surveyor.