Abstract
The rise of Industry 5.0 has shifted robotic systems toward human-centric design, emphasizing collaboration between humans and machines. While collaborative robots (cobots) are equipped with sensors for safety, they typically overlook human internal states such as fatigue, stress, and workload—factors that significantly impact performance. This lack of awareness limits their ability to adapt and collaborate effectively, particularly in dynamic, real-time environments where human conditions fluctuate. This thesis introduces a novel framework that integrates human internal states into robot decision-making to enhance human-robot collaboration. This research leverages reinforcement learning (RL) to develop methods that enable robots to adapt their behavior based on human physiological signals, enhancing team performance and interaction fluency. A key contribution is the exploration of offline RL techniques to address challenges in real-time human data collection, facilitating the training of human-aware RL agents in a safe and resource-efficient manner. Additionally, this work introduces Modality Utilization (MU) and State Utilization (SU) metrics—tools designed to quantify an RL agent’s reliance on different input features, including human data. These metrics enhance explainability and provide mechanisms to detect and mitigate over-reliance on any single information modality. By advancing human-aware robotic decision-making, this research contributes to the development of adaptive, collaborative systems aligned with Industry 5.0 principles.
Library of Congress Subject Headings
Human-robot interaction; Human engineering; Reinforcement learning
Publication Date
4-2025
Document Type
Dissertation
Student Type
Graduate
Degree Name
Electrical and Computer Engineering (Ph.D)
Department, Program, or Center
Electrical and Computer Engineering Technology
College
Kate Gleason College of Engineering
Advisor
Jamison Heard
Advisor/Committee Member
Ferat Sahin
Advisor/Committee Member
Eli Saber
Recommended Citation
Singh, Saurav, "Human-Aware Reinforcement Learning for Adaptive Human-Robot Teaming" (2025). Thesis. Rochester Institute of Technology. Accessed from
https://repository.rit.edu/theses/12186
Campus
RIT – Main Campus
Plan Codes
ECE-PHD
