Abstract
Bats easily identify targets of interest and reject clutter in the wild. A bat's transmit and receive beampatterns are frequency dependent, so it is possible that they exploit spectral cues to distinguish on-axis targets from off-axis clutter. The frequency dependent beampattern can also be interpreted as a different lowpass filter for each angular arrival. Given the known transmitted signal, the angle of arrival can be estimated from the high frequency attenuation in the received signal. The broadband Fisher information (FI) for azimuth in active sonar predicts a maximum FI at a slightly off-axis location, suggesting bats might target their sonar beam askance of a target to maximize the precision in estimating azimuth. This paper tests the FI model in a laboratory with a baffled tweeter as the sonar transmitter and an omnidirectional receiver. The experimentally measured beampatterns are used in Monte Carlo simulations with a maximum-likelihood estimator for azimuth over a range of SNR. The experimental results gained from the laboratory tests are compared with the model to assess the agreement between the predicted azimuth of maximum FI and the observed azimuth minimizing experimental error variance. [Funded by UMass Dartmouth Office of Undergraduate Research, College of Engineering, and ONR.]