Acoustic flow is the change in acoustic parameters in an acoustic scene induced by the relative motion between the observer (source) and the scene. This work investigates how bats may perceive acoustic flow through echolocation and use it to navigate. This is explored through both physical observations on bats’ response towards an induced acoustic flow, and computational simulations of how acoustic signals (specifically bat calls) behave in dynamic and complex acoustic scenes. The physical experiment studied whether free-flying wild pipistrelles were able to perceive a change in acoustic flow velocity by manipulating the motion of the acoustic background while the bats flew down a woodland corridor. It was found that the bats changed their speed proportionally to the relative velocity induced. To determine what flow information the bats may have processed, computational simulations were used to study acoustic flow in the context of bat flight. Frequency changes (or Doppler shift) proved to be the acoustic property that was most robust in estimating acoustic flow velocity. The simulations were then used to estimate flow induced frequency changes in complex environments and process the information for navigation. The Wideband Ambiguity processing method was used, which is a cross correlation of the echo signal with artificially Doppler shifted versions of the original signal (matched filtering) to extract the frequency changes. The simulation study manipulated different variables (signal properties, environment complexity and processing methods) to fully understand the limitations and potential of using Doppler to estimate flow velocity for bat-like signals. An algorithm was developed for processing a single signal-echo response to obtain lateral position as well as azimuth to a ‘wall’ of objects. The robustness of this algorithm for simple autonomous navigation scenarios, such as parallelly following a wall at a safe distance and turning a corner was then explored using computational simulations.
|Date of Award||23 Mar 2021|
- The University of Bristol
|Supervisor||Shane P Windsor (Supervisor) & Marc W Holderied (Supervisor)|