Deep Belief Net Learning in a Long-Range Vision System for Autonomous Off-Road Driving

被引:46
作者
Hadsell, Raia [1 ]
Erkan, Ayse [1 ]
Sermanet, Pierre [1 ,2 ]
Scoffier, Marco [2 ]
Muller, Urs [2 ]
LeCun, Yann [1 ]
机构
[1] NYU, Courant Inst Math Sci, New York, NY 10003 USA
[2] Net Scale Technol, Morganville, NJ USA
来源
2008 IEEE/RSJ INTERNATIONAL CONFERENCE ON ROBOTS AND INTELLIGENT SYSTEMS, VOLS 1-3, CONFERENCE PROCEEDINGS | 2008年
关键词
D O I
10.1109/IROS.2008.4651217
中图分类号
TP18 [人工智能理论];
学科分类号
081104 ; 0812 ; 0835 ; 1405 ;
摘要
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.
引用
收藏
页码:628 / 633
页数:6
相关论文
共 16 条
[11]  
LEIB D, 2005, P ROB SCI SYST RSS J
[12]  
RANZATO M, 2007, P C COMP VIS PATT RE
[13]  
SERMANET P, 2008, P INT C INT ROB SYST
[14]  
SOFMAN B, 2006, P ROB SCI SYST RSS J
[15]  
Stavens D., 2006, P C UNC AI UAI
[16]  
WELLINGTON C, 2004, P INT C ROB AUT ICRA