Hoverfly vision circuitry dramatically improves long-range drone detection
Consumer-grade drones can be distracting when in close proximity, but detecting them from a distance has been difficult, especially given their potential as weapons or surveillance tools, as well as their ability to stop people. helicopter and commercial flight operations.
New research from Flinders University and the University of South Australia has made an impressive leap in long-range drone detection, using signal processing algorithms inspired by the hoverfly’s ability to see targets in very low light environments.
Indeed, these insects are constantly working to select useful visual information from a large number of confusing noises, and as a result, they have developed photoreceptor systems that are extremely efficient at taking in a complex, cluttered and complex stream of information. noisy, and dramatically boost the signal-to-noise ratio to skip the useful bits.
The Flinders team set out to test if these sorts of photoreceptor circuits could do the same sort of thing with audio, taking a microphone array and analyzing ambient sound with a particular focus on amplification and detection of noise. sound of drones in flight.
They built a set of test microphones and connected them to a “biologically inspired vision” (BIV) processor designed to mimic the neural pathways of the hoverfly’s visual system. Processing the audio waveforms as two-dimensional “spectrogram” images, they passed the information through several stages of hover-inspired processing, including dynamic range reduction, adaptive filtering, pass-split feedback -low “DeVries-Rose”, low exponential “Weber” -pass divisional feedback and a nonlinear Naka-Rushton transform.
The results were very impressive. Compared to traditional broadband/narrowband drone detection systems, the BIV system was able to detect drones between 30% and 49% further away. The tests used drones as big as DJI’s Matrice 600 and as small as a Mavic Air, and the system was able to detect them between 3 and 4 kilometers (1.8 to 2.5 miles) in open country, and bring back more accurate estimates of flight parameters and trajectories.
“[The Hoverfly’s] ability to track their prey… Their early vision system is a very advanced motion identification system,” said Anthony Finn, professor of autonomous systems at UniSA. “This allows us to significantly improve the detectability of signatures.
A number of projects are considering this type of system to increase the low-light/noise signal capabilities of camera systems, but Finn says this is the first time a biovision system has been used with acoustic data. So while it’s impressive how much this increases the range of drone detection systems, it will also be very interesting to see what other audio applications could benefit from this kind of processing as research continues.
The document is freely available at The Journal of the Acoustical Society of America.
Watch a short video below.
Hoverfly brains mapped to detect acoustic signals from drones
Source: University of South Australia