TactileGCN: A Graph Convolutional Network forPredicting Grasp Stability with Tactile Sensors

Please use this identifier to cite or link to this item: http://hdl.handle.net/10045/90611
Información del item - Informació de l'item - Item information
Title: TactileGCN: A Graph Convolutional Network forPredicting Grasp Stability with Tactile Sensors
Authors: Garcia-Garcia, Alberto | Zapata-Impata, Brayan S. | Orts-Escolano, Sergio | Gil, Pablo | Garcia-Rodriguez, Jose
Research Group/s: Automática, Robótica y Visión Artificial | Robótica y Visión Tridimensional (RoViT) | Informática Industrial y Redes de Computadores
Center, Department or Service: Universidad de Alicante. Departamento de Física, Ingeniería de Sistemas y Teoría de la Señal | Universidad de Alicante. Departamento de Ciencia de la Computación e Inteligencia Artificial | Universidad de Alicante. Departamento de Tecnología Informática y Computación | Universidad de Alicante. Instituto Universitario de Investigación Informática
Keywords: Tactile detection | Tactile sensing | Tactile image | Artificial Intelligence | GCN | Graph Neural Network | Prediction of grasp stability
Knowledge Area: Ingeniería de Sistemas y Automática | Ciencia de la Computación e Inteligencia Artificial | Arquitectura y Tecnología de Computadores
Issue Date: 18-Jan-2019
Abstract: Tactile sensors provide useful contact data during the interaction with an object which can be used to accurately learn to determine the stability of a grasp. Most of the works in the literature represented tactile readings as plain feature vectors or matrix-like tactile images, using them to train machine learning models. In this work, we explore an alternative way of exploiting tactile information to predict grasp stability by leveraging graph-like representations of tactile data, which preserve the actual spatial arrangement of the sensor's taxels and their locality. In experimentation, we trained a Graph Neural Network to binary classify grasps as stable or slippery ones. To train such network and prove its predictive capabilities for the problem at hand, we captured a novel dataset of approximately 5000 three-fingered grasps across 41 objects for training and 1000 grasps with 10 unknown objects for testing. Our experiments prove that this novel approach can be effectively used to predict grasp stability.
Sponsor: This work has been funded by the Spanish Government with Feder funds (TIN2016-76515-R and DPI2015-68087-R), by two grants for PhD studies (FPU15/04516 and BES-2016-07829), by regional projects (GV/2018/022 and GRE16-19) and by the European Commission (COMMANDIA SOE2/P1/F0638), action supported by Interreg-V Sudoe.
URI: http://hdl.handle.net/10045/90611
Language: eng
Type: info:eu-repo/semantics/conferenceObject
Rights: © The authors
Peer Review: no
Publisher version: https://arxiv.org/abs/1901.06181
Appears in Collections:INV - RoViT - Comunicaciones a Congresos, Conferencias, etc.
INV - AUROVA - Comunicaciones a Congresos Internacionales
INV - I2RC - Comunicaciones a Congresos, Conferencias, etc.

Files in This Item:
Files in This Item:
File Description SizeFormat 
ThumbnailPID5818283.pdf4,38 MBAdobe PDFOpen Preview

Items in RUA are protected by copyright, with all rights reserved, unless otherwise indicated.