DocumentCode :
3099032
Title :
Deep belief net learning in a long-range vision system for autonomous off-road driving
Author :
Hadsell, Raia ; Erkan, Ayse ; Sermanet, Pierre ; Scoffier, Marco ; Muller, Urs ; LeCun, Yann
Author_Institution :
Courant Inst. of Math. Sci., New York Univ., New York, NY
fYear :
2008
fDate :
22-26 Sept. 2008
Firstpage :
628
Lastpage :
633
Abstract :
We present a learning-based approach for long-range vision that is able to accurately classify complex terrain at distances up to the horizon, thus allowing high-level strategic planning. A deep belief network is trained with unsupervised data and a reconstruction criterion to extract features from an input image, and the features are used to train a realtime classifier to predict traversability. The online supervision is given by a stereo module that provides robust labels for nearby areas up to 12 meters distant. The approach was developed and tested on the LAGR mobile robot.
Keywords :
belief networks; feature extraction; image classification; mobile robots; path planning; robot vision; strategic planning; unsupervised learning; LAGR mobile robot; autonomous off-road driving; deep belief net learning; deep belief network; features extraction; high-level strategic planning; long-range vision system; Convolutional codes; Distance measurement; Feature extraction; Meteorology; Navigation; Robots; Training;
fLanguage :
English
Publisher :
ieee
Conference_Titel :
Intelligent Robots and Systems, 2008. IROS 2008. IEEE/RSJ International Conference on
Conference_Location :
Nice
Print_ISBN :
978-1-4244-2057-5
Type :
conf
DOI :
10.1109/IROS.2008.4651217
Filename :
4651217
Link To Document :
بازگشت