Skip to Main Content
Sound source localization systems are typically made of free field microphone arrays. We have proposed the design of a localizing system based on the principles observed in nature, where directional acoustic sensing evolved to rely on diffraction about the head with only two ears. Localization is performed using the resultant frequency dependent acoustic phase and intensity differences between the two ears. These interaural functions can be computed analytically by modeling the head as a sphere. Now we report the first successful implementation of these ideas in an artificial head with two antipodally placed microphones. In order to determine source direction, we define a suitable metric associated to interaural functions. The true source direction is given by the global minimum of the metric between a measurement and a table of theoretical functions. The system was tested with a broadband source and was found to perform well even in a non-ideal reverberant laboratory environment. For comparison, we also calculated the source direction using the standard cross-correlation algorithm which is based on interaural time delay. Our approach performs more accurately and allows to use briefer signal durations. A key motivation for this work is to devise effective means to guide robotic navigation in environments with acoustic sources. The apparatus described in this paper is installed on a mobile robot in the laboratory.
Decision and Control, 2003. Proceedings. 42nd IEEE Conference on (Volume:6 )
Date of Conference: 9-12 Dec. 2003