The Role of Appearance in Mapping and Navigation
|Derik Schroeter||University of Oxford||[Home Page]|
Notice: hosted by Benoit Morisset
Date: 2008-10-10 at 10:00
Location: EJ228 (SRI E building) (Directions)
The use of appearance information in robotics has experienced a renaissance in recent years, largely due to significant advances in the machine vision community. In particular, the availability of fast and robust interest point detectors and descriptors has paved the way to ever more complex machine vision tasks being routinely integrated into robotic systems. This progress has led to the emergence of new and exciting opportunities while at the same time changing our notion of classic robotics challenges. In particular, appearance-based techniques developed in the computer vision domain have emerged as a valuable complement to state-of-the-art SLAM solutions. A prime example is the robust closing of large loops in a vehicles trajectory using an appearance-based visual loop-closing engine. The salient point here is that the data-association problem can be addressed without metric reasoning considering what things look like as opposed to where they appear to be. Further, combined with efficient machine learning techniques, appearance information is paving the way to ever more abstract semantic representations of sensor data an area of increasing importance for autonomous operations in complex outdoor environments.
This talk aims to illustrate the impact of appearance-based methods on robotics by outlining three distinct applications. The first part considers the creation of large-scale topological maps of an environment using a generative probabilistic approach. In the second part we describe an appearance-based method of navigating such a topological map. Finally, we describe the appearance-based extraction of semantic labels from sensor data of outdoor urban environments.
Please arrive at least 10 minutes early as you will need to sign in by following instructions by the lobby phone at Building E. SRI is located at 333 Ravenswood Avenue in Menlo Park. Visitors may park in the parking lots off Fourth Street. Detailed directions to SRI, as well as maps, are available from the Visiting AIC web page. There are two entrances to SRI International located on Ravenswood Ave. Please check the Builing E entrance signage.