In:
Adaptive Behavior, SAGE Publications, Vol. 20, No. 3 ( 2012-06), p. 159-171
Abstract:
Lizards have strong directional hearing across a broad band of frequencies. The directionality can be attributed to the acoustical properties of the ear, especially the strong acoustical coupling of the two eardrums. The peripheral auditory system of the lizard has previously been modeled in bilaterally symmetric form and several types of bilaterally asymmetric models have also been considered. Both in simulation by software and experiment with a mobile robot, the symmetric model shows a strong phonotaxis while asymmetric models show frequency-dependent direction errors determined by the type and magnitude of their intrinsic bias. To attain effective directional hearing, the bias in the peripheral system should be compensated. In this article, with the peripheral models, we design a decision model and a behavior model, a virtual robot, to simulate the auditory system of the lizard in software. The decision model is trained using three alternative training algorithms, which are evaluated and compared using experimental testing. The results demonstrate that the trained decision model is able to compensate for the frequency-dependent direction errors generated by the asymmetric peripheral models. With both symmetric and asymmetric peripheral models, the robot shows consistent and effective phonotaxis.
Type of Medium:
Online Resource
ISSN:
1059-7123
,
1741-2633
DOI:
10.1177/1059712312442402
Language:
English
Publisher:
SAGE Publications
Publication Date:
2012
detail.hit.zdb_id:
2070012-X
SSG:
12
SSG:
5,2
Permalink