In the past, several researchers focused on building accurate metric
or topological maps out of sensor data. The majority of approaches
present solutions to simultaneous localization and mapping but only
a few works try to acquire semantic information autonomously. In
this work we address the problem of classifying places in
environments into semantic classes based on range data only. We use
a supervised learning algorithm to train a set of classifiers based
on the Adaboost algorithm. Using our classification system, a mobile
robot is able to distinguish different places like rooms, corridors,
doorways, and hallways.
Thesis: [pdf: 1370k] [ps.gz: 1732k (better image quality)]
Bibtex
@mastersthesis{martinez2004thesis, title = {Supervised Learning of Places from Range Data using AdaBoost}, author = {Oscar Martinez Mozos}, school = {University of Freiburg}, month = {December}, year = {2004}, url = {http://www.informatik.uni-freiburg.de/~omartine/publications/thesis.pdf} }
Multimedia
Online classification with a mobile robot.
See how places are classified and colored as the robot moves.
Colors: room(blue), corridor(red), doorway(yellow).
Video: [avi: 77k].