Pragmatic Frames for Teaching and Learning in Human-Robot Interaction: Review and Challenges

Area of Science:

• Robotics and Human-Robot Interaction (HRI)
• Developmental Psychology
• Machine Learning

Background:

• Human-Robot Interaction (HRI) faces challenges in learning from inexperienced users, particularly for teaching skills flexibly.
• Leveraging human social competences and interactional mechanisms offers a path toward natural and efficient robot learning and teaching.
• Pragmatic frames, as flexible interaction protocols, provide crucial contextual cues for inferring and conveying skills.

Approach:

• This article defines and discusses pragmatic frames, drawing on developmental psychology research.
• It analyzes existing HRI literature on learning-teaching interactions, evaluating mechanisms in light of pragmatic frames.
• The study identifies limitations in current HRI applications of pragmatic frames compared to human-human interaction.

Key Points:

• Many HRI studies implicitly use basic elements of pragmatic frames for robot learning and teaching.
• Current HRI applications of pragmatic frames are restricted, hindering robust multi-task learning.
• Key absent features include rich, combinable repertoires of frames and continuous frame learning.

Conclusions:

• The restricted use of pragmatic frames in HRI impedes natural multi-task learning and teaching.
• Incorporating richer, combinable, and continuously learnable frames is crucial for advancing HRI.
• Future research should focus on solving challenges to fully leverage pragmatic frames for robot learning and teaching.