In this paper, we describe a preliminary experiment of citizenship engagement in the context of marine robotics using imitation learning to train a controller that mimics human behavior. The experiment has been carried out during the Festival della Comunicazione in Camogli, Italy, in September 2019. In more detail, citizens have been asked to pilot a small, light, and safe autonomous surface vehicle in front of a crowded public beach with the goal of performing an S-shaped path. The trajectories and controls performed by non-expert human operators have been recorded with the aim of training an imitation system that, after collecting a sufficient number of trajectory-control pairs, has been able to drive the vehicle without human intervention. To learn the human behavior, echo state networks have been employed as approximating architectures. The resulting controller turned out to be very effective in successfully performing the considered experiment with a reduced amount of training trajectories by imitating the human behavior also in unknown situations. The success of this experiment may pave the way to new research processes where citizens are actively engaged. Copyright (C) 2020 The Authors.

A preliminary experiment combining marine robotics and citizenship engagement using imitation learning

Odetti, Angelo;Bibuli, Marco;Cervellera, Cristiano;Ferretti, Roberta;Gaggero, Mauro;Zereik, Enrica;
2020

Abstract

In this paper, we describe a preliminary experiment of citizenship engagement in the context of marine robotics using imitation learning to train a controller that mimics human behavior. The experiment has been carried out during the Festival della Comunicazione in Camogli, Italy, in September 2019. In more detail, citizens have been asked to pilot a small, light, and safe autonomous surface vehicle in front of a crowded public beach with the goal of performing an S-shaped path. The trajectories and controls performed by non-expert human operators have been recorded with the aim of training an imitation system that, after collecting a sufficient number of trajectory-control pairs, has been able to drive the vehicle without human intervention. To learn the human behavior, echo state networks have been employed as approximating architectures. The resulting controller turned out to be very effective in successfully performing the considered experiment with a reduced amount of training trajectories by imitating the human behavior also in unknown situations. The success of this experiment may pave the way to new research processes where citizens are actively engaged. Copyright (C) 2020 The Authors.
File in questo prodotto:
File Dimensione Formato  
IFAC_WC_2020__SWAMP_Camogli.pdf

accesso aperto

Descrizione: Contributo in atti di convegno
Tipologia: Documento in Post-print
Dimensione 1.66 MB
Formato Adobe PDF
1.66 MB Adobe PDF Visualizza/Apri

I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.

Utilizza questo identificativo per citare o creare un link a questo documento: https://hdl.handle.net/11567/1098575
Citazioni
  • ???jsp.display-item.citation.pmc??? ND
  • Scopus 2
  • ???jsp.display-item.citation.isi??? 0
social impact