iCub Knows Where You Look: Exploiting Social Cues for Object Detection Learning

Posted by PaulHoule on Wednesday December 28, 2022 @12:15AM
from the arxiv.org dept.

( 2 comments | on HN )

Re:iCub Knows Where You Look: Exploiting Social Cues for Obj... by caramelazimuth on Thursday December 29, 2022 @07:44PM (#34180173)

Although in order for robots/agents to have properly meaningful, naturalistic interactions with humans they need to co-create a language of gaze and other joint attention cues:

https://link.springer.com/chapter/10.1007/978-3-031-19679-9_...

https://psyarxiv.com/ca8st/

[ Parent | on HN ]

Re:iCub Knows Where You Look: Exploiting Social Cues for Obj... by a9h74j on Thursday December 29, 2022 @09:13PM (#34180964)

I understand wanting to use gaze as a UX convenience. Has there been work done on human-robot interaction where a person sees a screen to "see" what the robot sees and can point and click to name/identify objects of interest?

[ Parent | on HN ]