ICDL 2021

The videos associated with our ICDL 2021 paper “A platform for embodied models of infant cognition, and its use in a model of event perception” by Mark Sagar, Alecia Moser, Annette Henderson, Sam Morrison, Nathan Pages, Alireza Nejati, Wan-Ting Yeh, Jonathan Conder, Alistair Knott, Khurram Jawed, Martin Takac (Soul Machines, Ltd; University of Auckland; Comenius University; University of Otago)

Paper abstract:

We are taking a holistic approach to modelling infant development. We are building BabyX – a computational model of ‘a whole baby’ including a graphical simulation of a baby’s physical body, autonomic functions (breathing, heartbeat), sensory systems (vision, hearing, touch, proprio/interoception), models of visual attention, visual object classification, category learning, reaching-to-grasp, visual object tracking, emotional system, focusing on subcortical pathways, neurochemicals and emotional behaviours.

In this paper we present our general model of event processing and describe how it processes simple motion events. We evaluate the motion event model using data from a real-baby experimental paradigm.

Sample Real Child – Parent/Caregiver interaction making a Farmyard Scene, showing the Parent/Caregiver’s view
Sample interaction scene involving event model in which BabyX observes and learns multimodally through self driven or user driven events while creating a farmyard scene, simulating real parent child interactions in our study.
Schematic display showing observation of events and generation of action events involving working memory while creating a farmyard scene.