PE-HRI: Multimodal Team Behaviors in Robot-Mediated Learning

34 teams of two children, aged 9-12, participated in the JUSThink robot-mediated learning activity. The dataset contains team-level multi-modal behavioral data, including interaction logs, speech, affective states, and gaze patterns, designed for studying productive engagement. It is grounded in engagement literature from Human-Robot Interaction and Intelligent Tutoring Systems.

Use Cases

• Validating theoretical models of engagement based on multi-modal behavioral features.
• Designing and evaluating robot pedagogical behaviors using logged interaction data.
• Analyzing collaboration patterns in child teams from speech and gaze data.
• Studying the relationship between affective states and learning outcomes in a constructivist activity.

Strengths

• Contains multi-modal data (log, speech, affect, gaze) for a holistic view of team behavior.
• Data is explicitly grounded in engagement literature from HRI and Intelligent Tutoring Systems.
• Collected from a controlled experimental setup with a specific pedagogical goal (minimum-spanning-tree problems).

Limitations

• Metadata completeness is low: specific column names, row counts, and file sizes are not provided in the sources.
• Dataset scale is limited to 68 participants (34 teams), which may constrain generalizability.
• Freshness is unknown as no last updated date is provided across the platforms.

Provenance

Source

Jauwairia Nasir, École Polytechnique Fédérale de Lausanne

Collection Method

Collected from a robot-mediated collaborative learning activity (JUSThink) involving children solving a minimum-spanning-tree problem in a gold mining scenario.

Geography

Experimental context involves a map of Switzerland, but the physical location of data collection is not specified.