Understanding Factors that Shape Children's Long Term Engagement with an In-Home Learning Companion Robot

Bengisu Çağıltay, Nathan J. White, Rabia Ibtasar, Bilge Mutlu, Joseph E. MichaelisarXiv 2022 · DOI · arXiv

This paper, presented at IDC '22, explores the "messy reality" of bringing social robots into the home. While we know robots can be great tutors in short bursts, the "novelty effect" usually wears off quickly. Researchers from UW-Madison and UIC set out to find what actually keeps a child engaged with a robot over a month-long period.

TL;DR

By deploying a reading companion robot in 14 homes for four weeks, researchers identified that long-term engagement is driven less by the robot’s "intelligence" and more by external factors like family routines, physical placement, and whether parental encouragement makes the interaction feel like a "chore."

What Problem Does This Paper Address?

The "Novelty Effect" is the bane of Human-Robot Interaction (HRI). Most studies happen in controlled labs or over very short durations. When a robot enters a real home, it has to compete with soccer practice, family vacations, and the allure of video games.

The authors address the gap in understanding sustained engagement. They want to know: What are the specific domestic and individual factors that cause a child to either adopt a robot into their daily life or abandon it after the first week?

Methods

The team designed a "Learning Companion" using the Misty II robot platform.

  • Hardware/Vision: The robot was paired with a Raspberry Pi 4. It used AprilTags (similar to QR codes) placed on the pages of 20 curated science books to "track" the child’s reading progress.
  • Interaction Design: As the child read aloud, the robot provided:
  • Knowledge Support: Summarizing text or making predictions.
  • Social Support: Expressing emotions or sharing "personal" stories (self-disclosure).
  • Interest Support: Connecting book themes to the child's life.
  • Study Flow: A 1-week baseline (reading without the robot) followed by a 4-week in-home deployment. The researchers used a "naturalistic" approach—families were free to use the robot as much or as little as they liked.
  • Analysis: They combined quantitative interaction logs (when and for how long the robot was used) with qualitative thematic analysis from seven semi-structured interviews per family.

Datasets

The study did not use pre-existing public datasets. Instead, it generated a rich, longitudinal dataset consisting of:

  • Interaction Logs: Time-stamped events of every robot encounter across 14 families.
  • Interview Transcripts: Qualitative data from 98 total interviews (7 per family).
  • Reading Journals: Logs of what books children read during the baseline and interaction weeks.

Key Insights and Ideas

  • The "Chore" Threshold: Parental involvement is a double-edged sword. While parents act as "prompters" to keep kids using the robot, if the prompting becomes too frequent, the child begins to view the robot as a "to-do list" item rather than a companion, leading to disengagement.
  • Physical Immediacy: Location is destiny. Robots placed in high-traffic areas (living rooms) saw more "spontaneous" use. If a robot was moved to a bedroom or a corner, it was often forgotten.
  • The "Non-Judgmental" Listener: Many children compared the robot to a pet or a dog. They felt comfortable reading aloud to it because, unlike a parent or teacher, the robot wouldn't judge them for mispronouncing a word.
  • Social vs. Informational Preferences: Children’s interests were highly idiosyncratic. Some loved the robot’s "social" chatter (e.g., Misty talking about her "robot friends"), while others found it repetitive and just wanted the science facts.

Key Results

The researchers categorized the children into four "engagement archetypes":

  1. Case 1: Modifiers: Children who kept using the robot but stopped reading aloud or showing tags, essentially "hacking" the interaction to suit their mood.
  2. Case 2: Discontinuers: Children who stopped using the robot entirely, often because they ran out of books they liked or found the social commentary repetitive.
  3. Case 3: Interrupted: Children whose engagement was killed by external "life events" (vacations, holidays) and never recovered.
  4. Case 4: Adopters: The 6 out of 14 children who successfully integrated the robot into a daily routine (e.g., before bed).

Quantitative Trend: Usage generally declined over the four weeks. In Week 1, 10 families used the robot "a few times a week." By Week 4, that number dropped to 7, and 4 families had stopped using it entirely.

Limitations and Future Work

  • Remote Constraints: Due to COVID-19, the study was fully remote. This meant researchers couldn't troubleshoot hardware issues in person, which led to two families dropping out early.
  • The "Researcher Effect": The weekly interviews likely acted as an artificial prompt, keeping some children engaged longer than they would have been otherwise.
  • Future Direction: The authors suggest that future robots need adaptive personalization. Instead of a "one-size-fits-all" social script, robots should use AI to detect when a child is getting bored and switch activities—moving from reading to games or "caretaking" (e.g., "feeding" the robot) to maintain the bond.

Summary updated

Abstract

Social robots are emerging as learning companions for children, and research shows that they facilitate the development of interest and learning even through brief interactions. However, little is known about how such technologies might support these goals in authentic environments over long-term periods of use and interaction. We designed a learning companion robot capable of supporting children reading popular-science books by expressing social and informational commentaries. We deployed the robot in homes of 14 families with children aged 10-12 for four weeks during the summer. Our analysis revealed critical factors that affected children's long-term engagement and adoption of the robot, including external factors such as vacations, family visits, and extracurricular activities; family/parental involvement; and children's individual interests. We present four in-depth cases that illustrate these factors and demonstrate their impact on children's reading experiences and discuss the implications of our findings for robot design.
BibTeX
@misc{cagltay2022understanding,
  author = {Bengisu Çağıltay and Nathan J. White and Rabia Ibtasar and Bilge Mutlu and Joseph E. Michaelis},
  title = {{Understanding Factors that Shape Children's Long Term Engagement with an In-Home Learning Companion Robot}},
  year = {2022},
  eprint = {2205.09052},
  archivePrefix = {arXiv},
  url = {https://arxiv.org/abs/2205.09052},
}
Figure