Vehicle Localization With A Downward Facing LIDAR

Open Access
Mattes, Richard J
Graduate Program:
Mechanical Engineering
Master of Science
Document Type:
Master Thesis
Date of Defense:
June 22, 2015
Committee Members:
  • Sean N Brennan, Thesis Advisor
  • localization
  • particle filter
  • autonomous vehicles
  • simulation
Autonomous, or self-driving vehicles hold much promise for the future of personal transportation. Autonomous vehicles have the potential to remove the human error from driving, reducing the occurrence of crashes and increasing the efficiency of traffic flow. One of the essential enabling technologies for autonomous vehicle operation is accurate and reliable vehicle position estimation within an environment, or vehicle localization. This thesis describes a particle filter based method for robust vehicle localization in a known environment using measurements of vehicle speed and reflected intensity measurements from a downward-facing scanning laser rangefinder (LIDAR). The approach was first tested in a proof-of-concept study in an idealized simulated environment. After showing that the method was able to reliably localize a vehicle in an ideal environment, it was then transitioned to data collected from a real vehicle traversing two separate environments. The results of the evaluation of the approach over the collected datasets are shown and discussed.