Towards Modeling Social-Cognitive Mechanisms in Robots to Facilitate Human-Robot Teaming

Abstract

For effective human-robot teaming, robots must gain the appropriate social-cognitive mechanisms that allow them to function naturally and intuitively in social interactions with humans. However, there is a lack of consensus on social cognition broadly, and how to design such mechanisms for embodied robotic systems. To this end, recommendations are advanced that are drawn from HRI, psychology, robotics, neuroscience and philosophy as well as theories of embodied cognition, dual process theory, ecological psychology, and dynamical systems. These interdisciplinary and multi-theoretic recommendations are meant to serve as integrative and foundational guidelines for the design of robots with effective social-cognitive mechanisms.

An interdisciplinary taxonomy of social cues and signals in the service of engineering robotic social intelligence

“…Here the capacity to embody one's own mental state and intentions is of import (i.e., specifying intentions through kinematic movements) given this allows humans, and ideally robots, to establish predictability, maintain shared representations, and facilitate interactions in a way that can maximize the success of joint tasks 20 . These issues, however, are just as theoretical as they are technical, and investigators are necessarily exploring these issues by integrating insights across disciplines [21][22][23][24] . While the implications of embodiment have allowed for the design of physical systems that can act dynamically within the niche of its physical environment 25 , functional performance within a social environment requires an understanding of embodiment as it applies to intelligent social behavior.…”

“…While recent works have attempted to integrate the socialcognitive theories underlying SSP into a cohesive framework from which models can be established [21][22][23][24] , little effort has been given toward incorporating the current body of knowledge regarding data-capture and the classification of signals and cues into a taxonomy capable of guiding researchers across disciplines 30 . An initial attempt at such a taxonomy, however, has been produced with respects to implementations within social robotics, providing a distinction between natural cues/signals (i.e., those present in humans that can be mimicked by robots) and artificial cues/signals (i.e., those that only an artificial agent can express, such as beeps, LED lights, etc.)…”

“…In sum, engineering social cognition requires an understanding of how signals are embodied through social cues comprising behavior and physiology in support of developing frameworks capable of endowing robots with their own social-cognitive systems or theory of mind 24,29 . We must consider innovative approaches toward making mechanical sensor systems better reflect the morphology and functioning of human sensory systems, and draft models in support of socio-cognitive architectures that can synthesize an adaptive response within real-world environments in real-time.…”

Clustering social cues to determine social signals: developing learning algorithms using the "n-most likely states" approach

“…We've also conducted work in synthesis through research on computational approaches for representing cues and their relationship to signals 17,18,19 , as well as how robots can be programmed to exhibit cues and test how that elicits particular interpretations (signals) from humans 3,20 . With this paper, we address a combination of the above, but include elements of analysis through the use of computational tools such as machine learning.…”

Me and My Robot Smiled at One Another: The Process of Socially Enacted Communicative Affordance in Human-Machine Communication

“…This body enables them to both perceive and act socially in their environment (Paauwe et al, 2015). Further, an agents' corporality is intrinsically related to how the body allows the robot not only to sense its environment and act in response to it, but also to exert an action as an agent in the environment (Wiltshire et al, 2013).…”