Abstract
One of the most difficult sensorimotor behaviors exhibited by flying animals is the ability to track another flying animal based on its sound emissions. From insects to mammals, animals display this ability in order to localize and track conspecifics, mate or prey. The pursuing individual must overcome multiple non-trivial challenges including the detection of the sounds emitted by the target, matching the input received by its (mostly) two sensors, localizing the direction of the sound target in real time and then pursuing it. All this has to be done rapidly as the target is constantly moving. In this project, we set to mimic this ability using a physical bio-mimetic autonomous drone. We equipped a miniature commercial drone with our in-house 2D sound localization electronic circuit which uses two microphones (mimicking biological ears) to localize sound signals in real-time and steer the drone in the horizontal plane accordingly. We focus on bat signals because bats are known to eavesdrop on conspecifics and follow them, but our approach could be generalized to other biological signals and other man-made signals. Using two different experiments, we show that our fully autonomous aviator can track the position of a moving sound emitting target and pursue it in real-time. Building an actual robotic-agent, forced us to deal with real-life difficulties which also challenge animals. We thus discuss the similarities and differences between our and the biological approach.
Highlights
Many animals are hypothesized to use real-time audio processing to localize and track conspecifics, mates, or prey [1–4]
We do so using a bio-inspired approach by developing a miniature electronic circuit with two ear-like microphones and a micro-processor that is placed on a miniature drone
The circuit detects ultrasonic signals that are typical for echolocating bats and it uses its two ‘ears’ to estimate the azimuth of the sound source and to steer the drone
Summary
Many animals are hypothesized to use real-time audio processing to localize and track conspecifics, mates, or prey [1–4]. Noteworthy are animals that track sounds emitted by other aviators who are themselves in flight These animals must apply rapid sensorimotor algorithms in realtime. Bats are hypothesized to move in groups in search of prey by eavesdropping on each other’s echolocation signals [9–12]. In all these cases, the aviators must detect the desired sounds and localize them and adapt their own flight-trajectories within dozens of milliseconds. The aviators must detect the desired sounds and localize them and adapt their own flight-trajectories within dozens of milliseconds Such animals require especially fast sensorimotor algorithms which are both challenging to reveal and potentially beneficial to mimic. Some of the specific challenges include detecting the correct sound signals within background noise, matching the sound signals arriving at the two ears to estimate source-direction and applying relevant algorithms to control the necessary flight maneuver
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
