An Improved Pedestrian Motion Tracking System Assisted by Machine Learning?

Yuming Chen,Wei Li,Zhuoyin Si
DOI: https://doi.org/10.1145/3352411.3352427
2019-01-01
Abstract:Pedestrian motion tracking based on the Micro-electromechanical inertial measurement unit (MEMS-IMU) possesses the unique advantages in terms of autonomy and anti-disturbance, which has been widely approved as a promising selection in indoor positioning. For the traditional pedestrian indoor positioning system, the accuracy of Zero-velocity interval detection and the stability of Kalman Filter (KF) are the key problem that restricting the development of indoor positioning system. To overcome the drawbacks of traditional Zero-velocity detection algorithm in robustness and adaptability, a novel Zero-velocity detection algorithm based on Support Vector Machine (SVM) is proposed. The construction of the SVM model based on the characteristics of pedestrian foot movement, and genetic algorithm is utilized to online optimizing the key parameters of the SVM. Then, incorporating the concept of strong tracking into the KF to improve the accuracy and stability of KF, a novel Zero Velocity Update (ZUPT) method based on the strong tracking KF (STKF) is presented. The experiment results show that the proposed method has better positioning accuracy and robustness than traditional indoor positioning method.
What problem does this paper attempt to address?