top of page

Fundamentals

Short guides on the ideas behind our sensors: how sound can image the world, and how we turn echoes into 3D pictures.

Download them to read, share or print.

Why We Learn an HRTF With a Small MLP

Why we learn the HRTF of a bat, or of a sonar, with a small neural network on Fourier features of the direction, and how that smooths it over the sphere and across frequency.

Acoustic Flow

When a sonar moves through a static world, every echo moves in sensor space. How that acoustic flow follows from the motion of the sensor, and what the echoes that break it tell us.

From Call to Image

From call to image: matched filter, beamforming, envelope filtering and CFAR detection in an eRTIS sensor, step by step on one simulated measurement.

Why LAEAP?

Why we draw sonar images, beampatterns and bat HRTFs on the Lambert azimuthal equal-area map instead of an azimuth/elevation grid, and what that map keeps and distorts.

Points, Walls and Corners

Points, smooth walls, corners and round pillars all look like bright spots in a single sonar image. The motion of the robot tells them apart.

Beyond Delay-and-Sum

Going beyond delay-and-sum: MVDR, MUSIC and spatial smoothing, why a single sonar ping is hard on them, and how sparse reconstruction works with an irregular array like eRTIS.

Fluttering Wings, Fluttering Echoes

A beating insect wing flashes back acoustic glints. How they show up in the echo of a long constant-frequency call and of a short FM sweep, as bats hear them.

Why Random Arrays?

Why the 32 microphones of an eRTIS look scattered at random: grating lobes, aperture and spacing, and how a random layout with a broadband call gets a sharp image without ghosts.

From One Bit at 4.5 MHz to 450 kHz

Our MEMS microphones deliver one bit at 4.5 MHz. How noise shaping works, and how CIC, FIR and IIR filters turn that into 450 kHz samples, measured on a real eRTIS sensor.

Multiplying Instead of Adding

Delay-multiply-and-sum and the coherence factor: how multiplying microphone signals instead of adding them narrows the beam and lowers the sidelobes of a small 32-microphone array.

Range or Doppler? Choosing the Call

Long tones, sweeps, hyperbolic sweeps and multisines compared with one tool, the ambiguity function: which call measures range, which measures velocity, and how several emitters can call at once.

Research & Collaboration

Cosys lab collaborates with academic and industrial partners on research in acoustic sensing, embedded systems and intelligent perception. Get in touch to discuss potential collaborations, projects or academic opportunities.

bottom of page