Reactive underwater object inspection based on artificial electric sense - Archive ouverte HAL Access content directly
Journal Articles Bioinspiration and Biomimetics Year : 2016

Reactive underwater object inspection based on artificial electric sense

Abstract

The weakly electric fish can perform complex cognitive tasks based on extracting information from blurry electric images projected from their immediate environment onto their electro-sensitive skin. In particular they can be trained to recognize the intrinsic properties of objects such as their shape, size and electric nature. They do this by means of original perceptual strategies that exploit the relations between the physics of a self generated electric field, their body morphology and the ability to perform specific movement termed Probing Motor Acts (PMA). In this article we artificially reproduce and combine these PMA to build an autonomous control strategy that allows an artificial electric sensor to find electrically contrasted objects and to orbit around them based on a minimum set of measurements and simple reactive feedback control laws of the probe's motion. The approach does not require any simulation model and could be implemented on an Autonomous Underwater Vehicle (AUV) equipped with artificial electric sense. The AUV has only to satisfy certain simple geometric properties, such as bilaterally (left/right) symmetrical electrodes and possess a reasonably high aspect (length/width) ratio.

Domains

Automatic
Fichier principal
Vignette du fichier
computational_morphology_BandB_form_V2.pdf (996.16 Ko) Télécharger le fichier
Origin : Files produced by the author(s)
Loading...

Dates and versions

hal-01503252 , version 1 (06-04-2017)

Identifiers

Cite

Vincent Lebastard, Frédéric Boyer, Sylvain Lanneau. Reactive underwater object inspection based on artificial electric sense. Bioinspiration and Biomimetics, 2016, 11, pp.45003 - 45003. ⟨10.1088/1748-3190/11/4/045003⟩. ⟨hal-01503252⟩
527 View
289 Download

Altmetric

Share

Gmail Facebook X LinkedIn More