Date of Award

12-2012

Document Type

Dissertation

Degree Name

Doctor of Philosophy (PhD)

Legacy Department

Electrical Engineering

Advisor

Hoover, Adam W

Committee Member

Brooks , Richard R

Committee Member

Gowdy , John N

Committee Member

Muth , Eric R

Abstract

The goal of this research is to improve the precision in tracking of an ultra-wideband (UWB) based Local Positioning System (LPS). This work is motivated by the approach taken to improve the accuracies in the Global Positioning System (GPS), through noise modeling and augmentation. Since UWB indoor position tracking is accomplished using methods similar to that of the GPS, the same two general approaches can be used to improve accuracy.
Trilateration calculations are affected by errors in distance measurements from the set of fixed points to the object of interest. When these errors are systemic, each distinct set of fixed points can be said to exhibit a unique set noise. For UWB indoor position tracking, the set of fixed points is a set of sensors measuring the distance to a tracked tag. In this work we develop a noise model for this sensor set noise, along with a particle filter that uses our set noise model. To the author's knowledge, this noise has not been identified and modeled for an LPS. We test our methods on a commercially available UWB system in a real world setting. From the results we observe approximately 15% improvement in accuracy over raw UWB measurements.
The UWB system is an example of an aided sensor since it requires a person to carry a device which continuously broadcasts its identity to determine its location. Therefore the location of each user is uniquely known even when there are multiple users present. However, it suffers from limited precision as compared to some unaided sensors such as a camera which typically are placed line of sight (LOS). An unaided system does not require active participation from people. Therefore it has more difficulty in uniquely identifying the location of each person when there are a large number of people present in the tracking area. Therefore we develop a generalized fusion framework to combine measurements from aided and unaided systems to improve the tracking precision of the aided system and solve data association issues in the unaided system. The framework uses a Kalman filter to fuse measurements from multiple sensors. We test our approach on two unaided sensor systems: Light Detection And Ranging (LADAR) and a camera system. Our study investigates the impact of increasing the number of people in an indoor environment on the accuracies using a proposed fusion framework. From the results we observed that depending on the type of unaided sensor system used for augmentation, the improvement in precision ranged from 6-25% for up to 3 people.

Share

COinS