A Machine Learning Approach to Understanding Older Adults' Trust in Assistive Robots

This project applies supervised machine learning models to predict older adults’ trust in novel assistive robots.

The research was conducted under the guidance of Dr. Samuel Olatunji and Dr. Wendy Rogers as an independent study in the Human Factors and Aging Laboratory at the University of Illinois Urbana-Champaign, and it was presented at the Illinois Undergraduate Research Symposium in April 2024.

Project Overview

We explored how factors such as prior technology experience, attitudes toward innovation, and demographics influence trust in assistive robots. The goal was to develop interpretable models that inform the design of technology better suited to the needs of aging populations.

Motivation

As the global population ages, assistive robots offer significant potential to support independence and quality of life. Understanding what drives trust in these technologies is essential to their successful adoption and inclusive design.

Methods

• Data Collection: Survey responses from adults aged 65+ about their trust in emerging assistive robots.
• Modeling Techniques: Lasso Regression and Ridge Regression.
• Evaluation Metrics: Mean Squared Error (MSE), R² Score, and K-Fold Cross-Validation.
• Features Used:
  • Familiarity: Self-reported familiarity with robot types across domains (e.g. autonomous cars, military robots, surgical robots, etc.)
  • Trust: Participant ratings of perceived robot trustworthiness (e.g. reliable, accurate, dependable, etc.)

Key Findings

• Prior familiarity with __________________ is a strong predictor of trust in assistive robots.

Presentations

• Illinois Undergraduate Research Symposium 2024 - Presentation Slides

Repo Structure

• data/: Deidentified familiarity and trust survey data from 3 studies
• notebooks/: Jupyter notebooks for EDA, modeling, visualization, and analysis
• figures/: Visuals for presentations and/or reports

Tech Stack

• Python
• Jupyter Notebooks
• scikit-learn
• pandas & matplotlib

Getting Started

1. Clone the repo
2. Install dependencies: pip install -r requirements.txt
3. Run notebooks in the notebooks/ folder using Jupyter Lab or Notebook
4. Explore results in the figures/ folder

License

Code in this repository is licensed under the MIT License.
Any included data, figures, or survey-derived content are provided under the
Creative Commons Attribution-NonCommercial 4.0 International (CC BY-NC 4.0) license unless otherwise noted.

Author

Saathveek Gowrishankar
Human Factors and Aging Laboratory, University of Illinois Urbana-Champaign
saathveek.com
LinkedIn
gsaathveek@gmail.com

Contact

Have questions, feedback, or want to collaborate? Feel free to reach out via email or connect with me on LinkedIn.