Autonomous Splash Detection System for Data Collection: A Subaqueous Multi-DAS Beamforming Approach

Grevenkop-Castenskiold, Bjørn Højmose ; Nielsen, Jonas Emil

4. semester

Electronic Systems, Master

2024

2024-05-31

128 pages

This report describes the development of a system capable of autonomously detecting, recording, and offloading splash events caused by people at bathing locations with the intent of creating a dataset to be used for training algorithms capable of confidently detecting potential drowning accidents in harbours. A system was designed, consisting of initial splash detection, calculation of direction of arrival (DOA) using time difference of arrival (TDOA), delay-and-sum (DAS) beamforming, confirmation of splash, recording and offloading data to a server. A subaqueous DAS beamformer was implemented on an ESP32-S3, consisting of three uniform linear arrays (ULAs), covering a frequency band each, to ensure suppression in the entire frequency band, receiving audio from 7 24-bit I2S microphones, impedance matched to water using nitrile gloves, over SPI from a PSoC 5LP, after which it was offloaded to a server using Wi-Fi. An automatic test system was implemented and used to verify the beampattern of the beamformer in air, by playing and recording an exponential sine sweep (ESS) and single tones at a time (STT) for each angle in an anechoic chamber. The measured beampatterns were then compared to simulations, where likeness was observed as well as some noise sources. The system was not evaluated in a watery setting, but is estimated to be ready for deployment with a few further developments and permit.

Beamforming ; Hydrophone ; DAS ; Differential ; ESP32 ; PSoC ; anechoic chamber ; drowning ; harbour ; safety ; detection ; autonomous ; dataset ; training data

Download
View record in AAU Student Projects