Fusion of the Integration and Parametric Approaches for PDR in Multiple Poses of Smartphone
Year: 2020
Type: dissertation
Abstract: In this dissertation, an IA-PA fusion-based PDR (Pedestrian Dead Reckoning) using low-cost inertial sensors is proposed to improve the indoor position estimation. Specifically, an IA (Integration Approach)-based PDR algorithm combined with measurements from PA (Parametric Approach) is constructed so that the algorithm is operated even in various poses that occur when a pedestrian moves with a smartphone indoors. In addition, I propose an algorithm that estimates the device attitude robustly in a disturbing situation by an ellipsoidal method. In addition, by using the machine learning-based pose recognition, it is possible to improve the position estimation performance by varying the measurement update according to the poses. \n \nFirst, I propose an adaptive attitude estimation based on ellipsoid technique to accurately estimate the direction of movement of a smartphone device. The AHRS (Attitude and Heading Reference System) uses an accelerometer and a magnetometer as measurements to calculate the attitude based on the gyro and to compensate for drift caused by gyro sensor errors. In general, the attitude estimation performance is poor in acceleration and geomagnetic disturbance situations, but in order to effectively improve the estimation performance, this dissertation proposes an ellipsoid-based adaptive attitude estimation technique. When a measurement disturbance comes in, it is possible to update the measurement more accurately than the adaptive estimation technique without considering the direction by adjusting the measurement covariance with the ellipsoid method considering the direction of the disturbance. In particular, when the disturbance only comes in one axis, the proposed algorithm can use the measurement partly by updating the other two axes considering the direction. The proposed algorithm shows its effectiveness in attitude estimation under disturbances through the rate table and motion capture equipment. \n \nNext, I propose a PDR algorithm that integrates IA and PA that can be operated in various poses. When moving indoors using a smartphone, there are many degrees of freedom, so various poses such as making a phone call, texting, and putting a pants pocket are possible. In the existing smartphone-based positioning algorithms, the position is estimated based on the PA, which can be used only when the pedestrian's walking direction and the device's direction coincide, and if it does not, the position error due to the mismatch in angle is large. In order to solve this problem, this dissertation proposes an algorithm that constructs state variables based on the IA and uses the position vector from the PA as a measurement. If the walking direction and the device heading do not match based on the pose recognized through machine learning technique, the position is updated in consideration of the direction calculated using PCA (Principal Component Analysis) and the step length obtained through the PA. It can be operated robustly even in various poses that occur. \n \nThrough experiments considering various operating conditions and paths, it is confirmed that the proposed method stably estimates the position and improves performance even in various indoor environments.
Author: 박소영
Language: English
Subfield: Electrical and Electronic Engineering
Field: Engineering
Domain: Physical Sciences
Open Access status: green
OpenAlex ID: W3138563677