Go to  Advanced Search

Indoor positioning through integration of optical angles of arrival with an inertial measurement unit

Show full item record

Files in this item

Files Size Format Description   View
ubc_2012_fall_islam_md. shariful.pdf 1.841Mb Adobe Portable Document Format   View/Open
Title: Indoor positioning through integration of optical angles of arrival with an inertial measurement unit
Author: Islam, Md. Shariful
Degree Master of Applied Science - MASc
Program Electrical and Computer Engineering
Copyright Date: 2012
Publicly Available in cIRcle 2012-09-24
Abstract: A novel indoor positioning solution is proposed in this work. An inertial navigation system (INS) is integrated with optical angle of arrival (OAOA) measurements to yield a smoother, more accurate, and robust positioning solution for indoor environments. An extended Kalman filter (EKF) is used to integrate the INS and OAOA measurements. Four different algorithms are proposed for the novel indoor positioning solution by INS/OAOA integration. An error state Kalman filter is used for implementing all four algorithms. In previous work, magnetometer error estimation was not included in the EKF state vector. In this work, magnetometer error estimation is added to the EKF state vector, and this reduced the average position error by 3.7% to 7%. Quaternion algebra is used instead of Euler angles due to the possibility of mathematical singularities for certain Euler angles. Quaternion vector estimation is performed by adding the quaternion vector to the state vector of the EKF. Both loosely coupled and tightly coupled integration strategies are explored for INS/OAOA integration. The tightly coupled strategy reduces the average positioning error by 60% compared to an OAOA-only system while the loosely coupled strategy reduces the average error by 44%. However, the performance improvement of the tightly coupled system comes with an increased computational cost due to nonlinearities in the measurement model. The loosely and tightly coupled algorithms are modified by augmenting the observation vector with a prior accelerometer bias estimate and a quaternion vector estimate. This results in loosely and tightly coupled algorithms with augmented observations. The algorithms with augmented observations perform significantly better, especially in a case of the low update rate for the OAOA sensor. An average position error of 4.89 cm is reduced to 3.11 cm by using the loosely coupled algorithm with augmented observations instead of the loosely coupled algorithm without observation augmentation. This is an improvement of approximately 36%. For the tightly coupled system, this improvement is approximately 32%. However, where the update rate from the OAOA sensor is fast enough, no significant performance improvement is observed by using algorithms with augmented observations.
URI: http://hdl.handle.net/2429/43261
Scholarly Level: Graduate

This item appears in the following Collection(s)

Show full item record

All items in cIRcle are protected by copyright, with all rights reserved.

UBC Library
1961 East Mall
Vancouver, B.C.
Canada V6T 1Z1
Tel: 604-822-6375
Fax: 604-822-3893