This thesis presents a comprehensive study of tactile sensing, particularly on the prob- lem of active texture perception. It includes a brief introduction to tactile sensing technology and the neural basis for tactile perception. It follows the literature review of textural percep- tion with tactile sensing. I propose a decoding and perception pipeline to tackle fine-texture classification/identification problems via active touching. Experiments are conducted using a 7DOF robotic arm with a finger-shaped tactile sensor mounted on the end-effector to per- form sliding/rubbing movements on multiple fabrics. Low-dimensional frequency features are extracted from the raw signals to form a perceptive feature space, where tactile signals are mapped and segregated into fabric classes. Fabric classes can be parameterized and sim- plified in the feature space using elliptical equations. Results from experiments of varied control parameters are compared and visualized to show that different exploratory move- ments have an apparent impact on the perceived tactile information. It implies the possibil- ity of optimising the robotic movements to improve the textural classification/identification performance.

Active Tactile Sensing for Texture Perception in Robotic Systems

WANG, SI AO
2023-07-27

Abstract

This thesis presents a comprehensive study of tactile sensing, particularly on the prob- lem of active texture perception. It includes a brief introduction to tactile sensing technology and the neural basis for tactile perception. It follows the literature review of textural percep- tion with tactile sensing. I propose a decoding and perception pipeline to tackle fine-texture classification/identification problems via active touching. Experiments are conducted using a 7DOF robotic arm with a finger-shaped tactile sensor mounted on the end-effector to per- form sliding/rubbing movements on multiple fabrics. Low-dimensional frequency features are extracted from the raw signals to form a perceptive feature space, where tactile signals are mapped and segregated into fabric classes. Fabric classes can be parameterized and sim- plified in the feature space using elliptical equations. Results from experiments of varied control parameters are compared and visualized to show that different exploratory move- ments have an apparent impact on the perceived tactile information. It implies the possibil- ity of optimising the robotic movements to improve the textural classification/identification performance.
27-lug-2023
robotics; tactile perception; artificial intelligence
File in questo prodotto:
File Dimensione Formato  
phdunige_4608134.pdf

accesso aperto

Tipologia: Tesi di dottorato
Dimensione 9.34 MB
Formato Adobe PDF
9.34 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/11567/1132355
Citazioni
  • ???jsp.display-item.citation.pmc??? ND
  • Scopus ND
  • ???jsp.display-item.citation.isi??? ND
social impact