Terrain-Based Road Vehicle Localization Using Attitude Measurements

Open Access
Dean, Adam
Graduate Program:
Mechanical Engineering
Doctor of Philosophy
Document Type:
Date of Defense:
October 30, 2008
Committee Members:
  • Sean Brennan, Dissertation Advisor
  • Sean N Brennan, Committee Chair
  • Henry Joseph Sommer Iii, Committee Member
  • Karl Martin Reichard, Committee Member
  • Jack W Langelaan, Committee Member
  • vehicle
  • terrain-based
  • estimation
  • kalman filter
  • particle filter
  • localization
This research is focused on developing a road vehicle localization technique that is independent of Global Positioning System (GPS) by means of a terrain-based method using in-vehicle attitude measurements. The motivation of this research is to improve vehicle autonomy, parameter estimation, and sensor redundancy by means of an accurate, drift-free vehicle localization algorithm. In order to enable several safety and efficiency features of road vehicle transportation, a drift-free localization algorithm that is independent of external signals or vision processes is developed. In-vehicle attitude measurements correlated to pre-recorded terrain maps are used to estimate vehicle position with sub-meter accuracy in most experiments. A particle filter algorithm is described in detail and used for estimating vehicle position. Several experiments are presented herein demonstrating the feasibility and various uses of this algorithm. These include longitudinal positioning along various types of roadways, lateral positioning or lane-indexing along multi-lane roadways, and lane departure in multi-path intersections. A method of approximating the steady-state estimate accuracy is also developed and experimentally validated. Results presented herein indicate the ability of pre-computing the sensor and mapping accuracy needed for achieving a desired threshold in positioning accuracy.