The ISL web site is maintained by
Most recent update 03/08/01.
Directional sensing of sound enables the localization of its source in
space. Many designed systems are comprised of free field
sensor arrays and use time of arrival differences betweeen
combinations of microphones to extract directional
information. In nature, directional acoustic sensing
evolved to rely on diffraction about the head with only two
sensors- the ears.
Inspired by human sound localization we have developed and implemented
an algorithm which compares measured interaural phase differences
(IPD) and interaural level differences (ILD) to theoretical
values to determine the direction of arrival. The algorithm and its
performance are described in the following two papers.
Through the use of MDLe
we have combined a sound-following atom which uses
our algorithm with an obstacle avoidance atom to create a
sound-following robot. Below we have several movies showing
the robot in action.
- A.A. Handzel and P.S. Krishnaprasad, "Biomimetic Sound-source
Localization," IEEE Sensors Journal,2(6):607-616,
- A.A. Handzel, S. Andersson, and P.S. Krishnaprasad, "A Biomimetic
Apparatus for Sound-source Localization," 41st IEEE Conf. on Decision and Control, to appear.
The current algorithm suffers from the "cone of confusion" common to
binaural systems; that is the localization yields no
information about the elevation of the sound source. In
current work we are investigating dynamical ideas to break
this symmetry. We have also used a broadband source; many
auditory sources in nature, however, are narrow band
(speech, for example). We are investigating the use of
duplex theory in which the information coming from the IPD
and ILD is weighted with respect to the frequency.
- Sound following demo - no obstacles
- Sound following demo - obstacle course, run 1
- Sound following demo - obstacle course, run 2
- Sound following demo - robot point of view