Perception of Localized Features During Robotic Sensorimotor Development

Alexandros Giagkos, Daniel Lewkowicz, Patricia Shaw, Suresh Kumar, Mark Lee, Qiang Shen

Research output: Contribution to journalArticlepeer-review

8 Citations (Scopus)
242 Downloads (Pure)

Abstract

The understanding of concepts related to objects are developed over a long period of time in infancy. This paper investigates how physical constraints and changes in visual perception impact on both sensorimotor development for gaze control, as well as the perception of features of interesting regions in the scene. Through a progressive series of developmental stages, simulating ten months of infant development, this paper examines feature perception toward recognition of localized regions in the environment. Results of two experiments, conducted using the iCub humanoid robot, indicate that by following the proposed approach a cognitive agent is capable of scaffolding sensorimotor experiences to allow gradual exploration of the surroundings and local region recognition, in terms of low-level feature similarities. In addition, this paper reports the emergence of vision-related phenomena that match human behaviors found in the developmental psychology literature
Original languageEnglish
Article number7843588
Pages (from-to)127-140
Number of pages14
JournalIEEE Transactions on Cognitive and Developmental Systems
Volume9
Issue number2
Early online date06 Feb 2017
DOIs
Publication statusPublished - 08 Jun 2017

Keywords

  • Biologically inspired feature extraction
  • conceptual learning through development
  • generation of representation during development
  • motor system and development
  • multimodal integration through development
  • robots with development and learning skills
  • visual system and development

Fingerprint

Dive into the research topics of 'Perception of Localized Features During Robotic Sensorimotor Development'. Together they form a unique fingerprint.

Cite this