SARC International Conference – Proceedings
Papers
-
A BIO-INSPIRED ROUTE TO CENTIMETER-SCALE NEAR-FIELD SOUND SOURCE LOCALIZATION FOR ROBOTS
MOHAMMAD NASR, HAMIDREZA NEMATI, EHSAN DEHGHAN-NIRI
Paper ID SA-MLAI-PTYA-240825-8186 · Pages 42–83
Download PDFAbstract
Robots increasingly need to pinpoint sounds within a few centimeters: manipulators that must identify the exact link struck in a collision,pipelineinspectors trying to localize the gas leakswith acoustic cameras and microphone arrays, and social robots that orient to a speaker’s mouth at arm’s length. Yet most robotic hearing still rests on far-field, plane-wave assumptions; when sources lie close, localization accuracy collapses or demands bulky microphone arrays. In this research, we propose a compact, bio-inspired near-field sound source localization (SSL)module modelled on the Aye-Aye (Daubentonia madagascariensis), a lemur that couples rapid finger taps withmobilepinnae to locate wood-boring larvae to sub-centimeter precision. Recently, we used a biomimetic approach by 3D printing aye-aye ear components to simulate this foraging technique. The findings indicate that the cupped pinna with the ear canal significantly improves acoustic field sensitivity by creating a focused area for sound reception and filtering out low frequencies to amplify higher, more crucial sounds. This specialized auditory system allows aye-ayes to efficiently detect subtle acoustic cues, highlighting an extraordinary adaptation for near-field sound detection.This architecture generalizes wherever centimeter-scale SSL matters: By utilizing Aye-Aye pinnae with controlled motion, we transform the usual near-field “nuisance†into a tightly focused sensing advantage, beneficial in automated pipeline maintenance, assembly and manipulation, and vision denied tasks. Keywords - Near-Field Sound Localization, Bio-inspired Robotics, Aye-Aye, Robot Audition.