Audio-Based Aircraft Detection System for Safe RPAS BVLOS Operations

Jorge Mariscal-Harana; Víctor Alarcón; Fidel González; Juan José Calvente; Francisco Javier Pérez-Grau; Antidio Viguria; Aníbal Ollero

doi:10.3390/electronics9122076

Audio-Based Aircraft Detection System for Safe RPAS BVLOS Operations

Jorge Mariscal-Harana, Víctor Alarcón, Fidel González, Juan José Calvente, Francisco Javier Pérez-Grau, Antidio Viguria, Aníbal Ollero

Biomedical Engineering Department

Research output: Contribution to journal › Article › peer-review

6 Citations (Scopus)

Abstract

For the Remotely Piloted Aircraft Systems (RPAS) market to continue its current growth rate, cost-effective ‘Detect and Avoid’ systems that enable safe beyond visual line of sight (BVLOS) operations are critical. We propose an audio-based ‘Detect and Avoid’ system, composed of microphones and an embedded computer, which performs real-time inferences using a sound event detection (SED) deep learning model. Two state-of-the-art SED models, YAMNet and VGGish, are fine-tuned using our dataset of aircraft sounds and their performances are compared for a wide range of configurations. YAMNet, whose MobileNet architecture is designed for embedded applications, outperformed VGGish both in terms of aircraft detection and computational performance. YAMNet’s optimal configuration, with >70% true positive rate and precision, results from combining data augmentation and undersampling with the highest available inference frequency (i.e., 10 Hz). While our proposed ‘Detect and Avoid’ system already allows the detection of small aircraft from sound in real time, additional testing using multiple aircraft types is required. Finally, a larger training dataset, sensor fusion, or remote computations on cloud-based services could further improve system performance.

Original language	English
Article number	2076
Pages (from-to)	1-13
Number of pages	13
Journal	Electronics
Volume	9
Issue number	12
DOIs	https://doi.org/10.3390/electronics9122076
Publication status	Published - 5 Dec 2020

Keywords

deep learning
sound event detection
convolutional neural networks
audio processing
embedded systems

Access to Document

10.3390/electronics9122076

https://www.mdpi.com/2079-9292/9/12/2076

Cite this

@article{9f356a51afee4336a0b63c9153f36fa3,

title = "Audio-Based Aircraft Detection System for Safe RPAS BVLOS Operations",

abstract = "For the Remotely Piloted Aircraft Systems (RPAS) market to continue its current growth rate, cost-effective {\textquoteleft}Detect and Avoid{\textquoteright} systems that enable safe beyond visual line of sight (BVLOS) operations are critical. We propose an audio-based {\textquoteleft}Detect and Avoid{\textquoteright} system, composed of microphones and an embedded computer, which performs real-time inferences using a sound event detection (SED) deep learning model. Two state-of-the-art SED models, YAMNet and VGGish, are fine-tuned using our dataset of aircraft sounds and their performances are compared for a wide range of configurations. YAMNet, whose MobileNet architecture is designed for embedded applications, outperformed VGGish both in terms of aircraft detection and computational performance. YAMNet{\textquoteright}s optimal configuration, with >70% true positive rate and precision, results from combining data augmentation and undersampling with the highest available inference frequency (i.e., 10 Hz). While our proposed {\textquoteleft}Detect and Avoid{\textquoteright} system already allows the detection of small aircraft from sound in real time, additional testing using multiple aircraft types is required. Finally, a larger training dataset, sensor fusion, or remote computations on cloud-based services could further improve system performance.",

keywords = "deep learning, sound event detection, convolutional neural networks, audio processing, embedded systems",

author = "Jorge Mariscal-Harana and V{\'i}ctor Alarc{\'o}n and Fidel Gonz{\'a}lez and Calvente, {Juan Jos{\'e}} and P{\'e}rez-Grau, {Francisco Javier} and Antidio Viguria and An{\'i}bal Ollero",

year = "2020",

month = dec,

day = "5",

doi = "10.3390/electronics9122076",

language = "English",

volume = "9",

pages = "1--13",

journal = "Electronics",

issn = "2079-9292",

publisher = "Multidisciplinary Digital Publishing Institute (MDPI)",

number = "12",

}

TY - JOUR

T1 - Audio-Based Aircraft Detection System for Safe RPAS BVLOS Operations

AU - Mariscal-Harana, Jorge

AU - Alarcón, Víctor

AU - González, Fidel

AU - Calvente, Juan José

AU - Pérez-Grau, Francisco Javier

AU - Viguria, Antidio

AU - Ollero, Aníbal

PY - 2020/12/5

Y1 - 2020/12/5

N2 - For the Remotely Piloted Aircraft Systems (RPAS) market to continue its current growth rate, cost-effective ‘Detect and Avoid’ systems that enable safe beyond visual line of sight (BVLOS) operations are critical. We propose an audio-based ‘Detect and Avoid’ system, composed of microphones and an embedded computer, which performs real-time inferences using a sound event detection (SED) deep learning model. Two state-of-the-art SED models, YAMNet and VGGish, are fine-tuned using our dataset of aircraft sounds and their performances are compared for a wide range of configurations. YAMNet, whose MobileNet architecture is designed for embedded applications, outperformed VGGish both in terms of aircraft detection and computational performance. YAMNet’s optimal configuration, with >70% true positive rate and precision, results from combining data augmentation and undersampling with the highest available inference frequency (i.e., 10 Hz). While our proposed ‘Detect and Avoid’ system already allows the detection of small aircraft from sound in real time, additional testing using multiple aircraft types is required. Finally, a larger training dataset, sensor fusion, or remote computations on cloud-based services could further improve system performance.

AB - For the Remotely Piloted Aircraft Systems (RPAS) market to continue its current growth rate, cost-effective ‘Detect and Avoid’ systems that enable safe beyond visual line of sight (BVLOS) operations are critical. We propose an audio-based ‘Detect and Avoid’ system, composed of microphones and an embedded computer, which performs real-time inferences using a sound event detection (SED) deep learning model. Two state-of-the-art SED models, YAMNet and VGGish, are fine-tuned using our dataset of aircraft sounds and their performances are compared for a wide range of configurations. YAMNet, whose MobileNet architecture is designed for embedded applications, outperformed VGGish both in terms of aircraft detection and computational performance. YAMNet’s optimal configuration, with >70% true positive rate and precision, results from combining data augmentation and undersampling with the highest available inference frequency (i.e., 10 Hz). While our proposed ‘Detect and Avoid’ system already allows the detection of small aircraft from sound in real time, additional testing using multiple aircraft types is required. Finally, a larger training dataset, sensor fusion, or remote computations on cloud-based services could further improve system performance.

KW - deep learning

KW - sound event detection

KW - convolutional neural networks

KW - audio processing

KW - embedded systems

UR - http://www.scopus.com/inward/record.url?scp=85097445971&partnerID=8YFLogxK

U2 - 10.3390/electronics9122076

DO - 10.3390/electronics9122076

M3 - Article

SN - 2079-9292

VL - 9

SP - 1

EP - 13

JO - Electronics

JF - Electronics

IS - 12

M1 - 2076

ER -

Audio-Based Aircraft Detection System for Safe RPAS BVLOS Operations

Abstract

Keywords

Access to Document

Other files and links

Fingerprint

Cite this