Positional estimation of invisible drone using acoustic array with A-shaped neural network

Jongsik Ahn, Min Young Kim

Research output: Chapter in Book/Report/Conference proceedingConference contributionpeer-review

1 Scopus citations

Abstract

Image-based object detection is a commonly used algorithm for anti-drone surveillance system. However, there is a disadvantage that it cannot be detected if the target is not visible within the image. In this paper, we propose drone position estimation algorithm using acoustic array to detect objects complementing the difficulty of estimating sudden directional shifts in hiding, occurrence situations and quickly out of the vision of the camera. Sound data is converted into an image via mel-spectrogram to facilitate image sensor and sound sensor fusion and the drone position is estimated via the Convolution Neural Network. The proposed neural network is the A-shape neural network, which consists of up-sampling and down-sampling. Through these methods, we achieve RMSE of 13.045 pixels and show that the location of the drone can be estimated efficiently.

Original languageEnglish
Title of host publication3rd International Conference on Artificial Intelligence in Information and Communication, ICAIIC 2021
PublisherInstitute of Electrical and Electronics Engineers Inc.
Pages320-324
Number of pages5
ISBN (Electronic)9781728176383
DOIs
StatePublished - 13 Apr 2021
Event3rd International Conference on Artificial Intelligence in Information and Communication, ICAIIC 2021 - Jeju Island, Korea, Republic of
Duration: 13 Apr 202116 Apr 2021

Publication series

Name3rd International Conference on Artificial Intelligence in Information and Communication, ICAIIC 2021

Conference

Conference3rd International Conference on Artificial Intelligence in Information and Communication, ICAIIC 2021
Country/TerritoryKorea, Republic of
CityJeju Island
Period13/04/2116/04/21

Keywords

  • Acoustic
  • Anti-Drone System
  • Convolution Neural Network
  • Mel-Spectrogram
  • Surveillance System

Fingerprint

Dive into the research topics of 'Positional estimation of invisible drone using acoustic array with A-shaped neural network'. Together they form a unique fingerprint.

Cite this