Consiglio Nazionale delle Ricerche

Tipo di prodottoArticolo in rivista
TitoloRecurrent and Convolutional Neural Networks for Deep Terrain Classification by Autonomous Robots
Anno di pubblicazione2020
Formato-
Autore/iFabio Vulpi (1,2), Annalisa Milella (2), Roberto Marani (2), Giulio Reina (1)
Affiliazioni autori(1) Department of Mechanics, Mathematics & Management, Polytechnic of Bari, Via Orabona 4, 70125, Bari, Italy (2) Institute of Intelligent Industrial Technologies and Systems for Advanced Manufacturing, National Research Council, via G. Amendola 122/O, 70126 Bari, Italy
Autori CNR e affiliazioni
  • FABIO VULPI
  • ANNALISA MILELLA
  • ROBERTO MARANI
Lingua/e
  • inglese
AbstractThe future challenge for field robots is to increase the level of autonomy towards long distance (>1 km) and duration (>1 h) applications. One of the key technologies is the ability to accurately estimate the properties of the traversed terrain to optimize onboard control strategies and energy efficient path-planning, ensuring safety and avoiding possible immobilization conditions that would lead to mission failure. Two main hypotheses are put forward in this research. The first hypothesis is that terrain can be effectively detected by relying exclusively on the measurement of quantities that pertain to the robot-ground interaction, i.e., on proprioceptive signals. Therefore, no visual or depth information is required. Then, artificial deep neural networks can provide an accurate and robust solution to the classification problem of different terrain types. Under these hypotheses, sensory signals are classified as time series directly by a Recurrent Neural Network or by a Convolutional Neural Network in the form of higher-level features or spectrograms resulting from additional processing. In both cases, results obtained from real experiments show comparable or better performance when contrasted with standard Support Vector Machine with the additional advantage of not requiring an a priori definition of the feature space.
Lingua abstractinglese
Altro abstract-
Lingua altro abstract-
Pagine da-
Pagine a-
Pagine totali-
RivistaJournal of terramechanics (Print)
Attiva dal 1964
Editore: Pergamon Press - Oxford
Paese di pubblicazione: Regno Unito
Lingua: inglese
ISSN: 0022-4898
Titolo chiave: Journal of terramechanics (Print)
Titolo proprio: Journal of terramechanics. (Print)
Titolo abbreviato: J. terramechanics (Print)
Numero volume della rivista-
Fascicolo della rivista-
DOI-
Verificato da referee-
Stato della pubblicazionePreprint
Indicizzazione (in banche dati controllate)-
Parole chiaveAutonomous robots, vehicle-terrain interaction, terrain classification, deep-learning
Link (URL, URI)-
Titolo parallelo-
Licenza-
Scadenza embargo-
Data di accettazione10/12/2020
Note/Altre informazioni-
Strutture CNR
  • STIIMA — Istituto di Sistemi e Tecnologie Industriali Intelligenti per il Manifatturiero Avanzato
Moduli/Attività/Sottoprogetti CNR-
Progetti Europei-
Allegati