Model-based Inclusion of Previewed Information for Lateral Vehicle State and Environment Estimation

Open Access
Author:
Brown, Alexander Allen
Graduate Program:
Mechanical Engineering
Degree:
Doctor of Philosophy
Document Type:
Dissertation
Date of Defense:
July 05, 2013
Committee Members:
  • Sean Brennan, Dissertation Advisor
  • Sean Brennan, Committee Chair
  • Henry Joseph Sommer Iii, Committee Member
  • Christopher Rahn, Committee Member
  • Jacob Willem Langelaan, Special Member
  • Richard Laurence Tutwiler, Special Member
Keywords:
  • MODEL-BASED ESTIMATION
  • PREVIEW ESTIMATION
  • STATE ESTIMATION
  • VEHICLE DYNAMICS
  • AUTONOMOUS VEHICLES
Abstract:
This dissertation is concerned with the marriage of spatiotemporal preview as it is known in the controls literature with map-based guidance in a simple but effective paradigm, in the interest of making accurate environment and vehicle states accessible with production-grade sensing and computing equipment. It does so without the two-dimensional data association issues of SLAM. Specifically, ideas from linear optimal preview control theory will be employed to develop mathematically tractable and intuitively insightful estimation frameworks that use map, inertial, and forward-looking monocular camera information. The framework developed is intended to provide state and path estimates to an optimal preview controller guiding a vehicle in a static environment at highway speeds. Three variations on preview-based estimation are considered, with experimental results for both human-driven and self-steering vehicles.