FDIO: Frequency Decomposed Inertial Odometry
Researchers have developed a new method called Frequency Decomposed Inertial Odometry (FDIO) to improve pedestrian motion tracking using only IMU data. FDIO addresses the challenge of separating human body motion from limb perturbations by decomposing IMU signals into low and high frequency components. A Mamba module processes the low-frequency data for long-range motion, while a multi-scale convolution module handles the high-frequency data for fine-grained dynamics. This approach has demonstrated significant error reductions compared to existing baselines on public datasets. AI
IMPACT This novel approach to inertial odometry could enhance the accuracy of motion tracking in consumer applications and robotics.