Emotion Pattern Recognition in Robotics and Emotionally Adaptive Robotics Prototype

Tufts HRI Lab Research • 2023–2024

Overview

This project explored whether emotionally adaptive robots can reduce stress in students more effectively than scripted robots. We developed a prototype that recognized stress patterns and adapted its interactions in real time, using the Misty II robot as the testing platform.

Problem

Students at Tufts reported high stress levels and faced long wait times for counseling services. The challenge was: could an emotionally adaptive robot provide immediate support and help reduce stress?

Approach

Two interaction modes: Scripted (Direct Selection) vs Emotionally Adaptive. The scripted mode delivered mindfulness instructions, while the adaptive mode checked in with users and adjusted responses dynamically.

Crossover study with 30 students: Each participant experienced both modes in randomized order, ensuring a fair comparison.

Therapy activity: Used a DBT-based mindfulness exercise, delivered by the Misty II robot programmed with Python and the MistyPy SDK.

Data collection: Pre- and post-interaction surveys measured stress levels, analyzed with RStudio to validate results.

Outcomes & Impact

My Role

I contributed to experiment design, participant recruitment, survey creation, programming the Misty II robot, data analysis in RStudio, and presenting results. My role bridged technical implementation with human-centered research insights.

Technologies Used

PythonROSMisty II SDKMachine LearningHuman-Robot Interaction

Research Paper

Reflection

This project reinforced core product management skills: defining clear problems, validating solutions with users, and turning technical insights into actionable outcomes. It highlighted the importance of empathy and user-centered design in robotics and future technology products.