Event Location: NSH 1305
Bio: Hernan Badino received his PhD degree from the J. W. Goethe Frankfurt University, in 2008. Dr. Badino has worked on vision based environment perception for driver assistance systems during his PhD at Daimler AG, in Stuttgart, Germany. He joined the Robotics Institute in 2009 as a post doctoral researcher where he worked on multi-sensor fusion, structure from motion, real-time object segmentation and classification, lane detection, and vehicle localization. He is now at NREC working on agricultural robotics.

Abstract: One of the fundamental requirements of an autonomous vehicle is the ability to determine its location on a map. Frequently, solutions to this localization problem rely on GPS information or use expensive 3D sensors. In this talk, I describe a method for long-term vehicle localization based on visual features alone. My approach utilizes a combination of topological and metric mapping, to encode the coarse topology of the route as well as detailed metric information required for accurate localization. A topometric map is created by driving the route once and recording a database of visual features. The vehicle then localizes by matching features to this database at runtime. Since individual feature matches are unreliable, I employ a discrete Bayes filter to estimate the most likely vehicle position along the route.
The algorithm is reliable indoors and outdoors and across wide environmental changes, including lighting differences, seasonal variations, and occlusions, achieving an average localization accuracy of 1 m over an 8 km route. The method runs in real time and converges correctly even with wrong initial position estimates solving the kidnapped robot problem.