Automatic identification of sound source position employing neural networks and rough sets
Pattern Recognition Letters - Special issue: Rough sets, pattern recognition and data mining
CVPR '06 Proceedings of the 2006 IEEE Computer Society Conference on Computer Vision and Pattern Recognition - Volume 1
Advanced sensorial system for an acoustic LPS
Microprocessors & Microsystems
| Hi-index | 0.00 |
Bearing-only Simultaneous Localization and Mapping (SLAM) is apartially observable SLAM problem, in wich the sensor used forperceiving the robot`s enviroment, provides only-angularinformation respect to the landmarks, and therefore does not giveenough information to compute the full state of a landmark from asingle observation. In this context, vision-based systems have alsogained a great interest in the robotics community. Nevertheless theuse of "sound sources" as map's features have been very littleexplored in SLAM. In this work a method for performing SLAM withsound sources is presented. A robot capable of sense bearinginformation respect to an external sound source with modest angularacuity (-10°) is considered.At the robot trajectory start, thesound source position is unknown; while the robot moves, theposition of the sound source and the robot position in a globalcoordinate frame are both estimated. Experimental results withsimulations and with a real robot demonstrate that tracking aunique source sound is enough to reasonably correct the odometryinformation provided by the encoders.