ETD

Archivio digitale delle tesi discusse presso l'Università di Pisa

Tesi etd-10282017-221538


Tipo di tesi
Tesi di dottorato di ricerca
Autore
PIERONI, MICHAEL
URN
etd-10282017-221538
Titolo
Bioinspired Artificial Vision
Settore scientifico disciplinare
ING-INF/06
Corso di studi
INGEGNERIA
Relatori
tutor Prof. De Rossi, Danilo
commissario Prof. Carpi, Federico
commissario Prof. Scilingo, Enzo Pasquale
commissario Prof. Viollet, Stephane
Parole chiave
  • Optics
  • Image Formation Simulation
  • Bioinspiration
  • Computer Vision
  • Artificial Vision
  • Animal Eyes
  • Social Robots
Data inizio appello
11/11/2017
Consultabilità
Non consultabile
Data di rilascio
11/11/2087
Riassunto
Bioinspired Arti cial Vision is a multi-disciplinary research eld aimed at developing technology and system implementing visual functions taking inspiration from studies on biological vision apparatus able to accomplish the same function. Such a type of arti cial systems include a visual body and a visual brain. The body is de ned as the whole of the robotic eyes optical and imaging components, and their associated optical and imaging early (low-level)processing, and the brain is the whole cognitive (high-level) visual processing. This PhD thesis targets three foundational innovations in all the phases of arti cial vision system design. First, a bioinspired technology to accomplish a low-level visual mechanism such as keep objects in focus. It consists in the design, development and characterization of a tunable lenses entirely made of soft solid matter (elastomers) whose working principle is inspired by the accommodation in bird and reptile. Second, a visual perception system which implement the high-level visual tasks (e.g. estimation of visually salient stimuli, face/gesture detection and recognition, etc.) that a humanoid robot need to extract during the social interaction with humans. Third and last innovation target a evolutionary-driven methodology in design the over-all arti cial vision system. State-of-the-art evolutionary algorithms have been used only in developing the visual brain through simulation. Here, I introduce a software tool that enable a realistic estimation of the visual body imaging response that can be used to model this part of the system in a simulated scenario
File