Graduation Year

2021

Document Type

Dissertation

Degree

Ph.D.

Degree Name

Doctor of Philosophy (Ph.D.)

Degree Granting Department

Electrical Engineering

Major Professor

Wilfrido A. Moreno, Ph.D.

Committee Member

Chung Seop Jeong, Ph.D.

Committee Member

Zhenyu Wang, Ph.D.

Committee Member

Mahshid Rahnamay Naeini, Ph.D.

Committee Member

Fernando Falquez, Ph.D.

Keywords

Kalman Filter, SE, SLAM, V2V, V2X

Abstract

Research and development in Connected Vehicles (CV) Technologies has increased exponentially, with the allocation of 75 MHz radio spectrum in the 5.9 GHz band by the Federal Communication Commission (FCC) dedicated to Intelligent Transportation Systems (ITS) in 1999 and 30 MHz in the 5.9 GHz by the European Telecommunication Standards Institution (ETSI). Many applications have been tested and deployed in pilot programs across many cities all over the world.

CV pilot programs have played a vital role in evaluating the effectiveness and impact of the technology and understanding the effects of the applications over the safety of road users. The evaluation of applications from the pilot program has resulted in the core interest in discussing the challenges CV technology faces. In CVs, the vehicle's position is monitored by Global Positioning System (GPS) via Global Navigation Satellite System (GNSS) antennas placed on the vehicle. While the precision of GPS presented by GPS.gov states that the accuracy of high-quality GPS has a horizontal accuracy of less than or equal to 1.8 meters 95% of the time. In comparison, GPS-enabled smartphones with built-in Inertial Measurement Units (IMU) have shown an accuracy of 4.9 meters [1] under open skies. Also, it is stated that the GPS accuracy is impacted by the interference of physical objects such as “buildings, bridges, trees, etc.”[2].

The current CV applications are predominantly dependent on the GPS for the vehicle’s location measurements, making the accuracy of the GPS a significant factor impacting the performance of the CV technology. For example, the average width of a car is about 1.65 meters, and the width of the lane ranges from 2.5 meters to 3.25 meters. Considering an estimated position error from the GPS being 1.8 meters, the car location could be estimated to be a lane away from its actual lane position.

In this research, a novel approach of coupling vehicular sensors to the GPS devised using adaptive algorithms is proposed. The simulations results support the proposed novel approach that can be adapted in CVs to increase vehicle localization accuracy. The results presented show the Root Mean Square Deviation (RMSD) of about 1.5 meters compared to the GPS values.

Using the Systems Engineering (SE) approach, the aim of this research topic is to contribute to increase the accuracy in CV localization. A conceptual design to integrate in vehicle sensors into CV technology is presented. A Python program framework to implement the sensor fusion techniques is contributed as a byproduct of the research. Using the framework analysis of the simulation results are presented and documented. Finally, RMSD is used to evaluate the goodness of the novel approach and a cost benefit analysis is provided to understand the benefit of the novel technique.

Download

Share

COinS