Bolita

Many children struggle with initiating conversations and building social confidence. Telling jokes can be an effective tool for breaking the ice, but children often lack the practice and confidence to do so successfully.

Ivonne Monarch, Researcher

July 2019- September 2020

bolita.mp4

Project Overview

Bolita is an innovative hybrid conversational agent designed to help children practice joke-telling in a supportive, judgment-free environment. By combining a Sphero robot with a Google Home assistant, we created a unique platform that:

Assists children in learning and practicing joke delivery
Provides a safe space for social skill development
Explores potential language interaction challenges

Key Findings

User Experience

Children showed high engagement with Bolita
Positive feedback across multiple experience dimensions

Research Goals

Investigate how children interact with voice assistance technologies when telling jokes
Analyze vocal production features during joke-telling interactions
Understand the impact of conversational agents on children's communication skills

Methodology

Design Approach

User-Centered Design: Conducted participatory design sessions with children and specialists
Joke database: Developed a database of short, child-friendly jokes reviewed by a psychologist
Interaction Design: Created a three-phase interaction model
1. Introduction
2. Joke Telling
3. Joke Repetition

Technical Implementation

Technologies: Dialogflow, Python
Hardware: Sphero Robot with LED matrix
Interaction Features:
- Equalizer animation during speech
- Laughing and rolling robot response

58% ♥️

Loved to speak with Bolia

44.6% 💬

Would like to speak with Bolita again

50% 😂

Found it super fun to speak with Bolita

80% 🏠

Would like to have Bolita in their homes

Technical Insights

Speech Recognition Challenges:
- Children with below-average language skills had higher Word Error Rates (WER)
  - Below-average language group: 25% WER
  - Other children: <10% WER
Voice Interaction Limitations:
- Current robotic voice may inadvertently encourage unnatural speech mimicry

Impact and Implications

Demonstrated potential of conversational agents in children's social skill development
Identified critical considerations for designing child-friendly voice interfaces
Highlighted importance of adaptive speech recognition for diverse language abilities

Future Directions

Improve voice naturalness
Develop more sophisticated speech recognition
Explore broader applications in child-centered communication technologies

Publications

Monarca et al., 2020 (Full research details available)

Personal Contribution

Led user-centered design process
Developed interaction flow and joke database
Conducted research implementation and data analysis
Developed technical infrastructure using Dialogflow and Python

Toolkit and Technologies

Dialogflow
Python
Sphero Robot
Google Home Assistant
Google Firebase

Related Publications:

Monarca, I., Cibrian, F. L., Mendoza, A., Hayes, G., & Tentori, M. (2020, September). Why doesn't the conversational agent understand me? a language analysis of children speech. In Adjunct Proceedings of the 2020 ACM International Joint Conference on Pervasive and Ubiquitous Computing and Proceedings of the 2020 ACM International Symposium on Wearable Computers (pp. 90-93).

Back