Caregivers’ Perception of Robot Errors

Current Project

As assistive robots enter caregiving spaces, it becomes critical to understand how caregivers interpret and respond to robot failures. My honors thesis investigates how formal (e.g., clinical professionals) and informal (e.g., family members) caregivers perceive robot errors during care tasks, and how their trust, intervention strategies, and expectations differ.

I designed a controlled lab study where participants interact with a robot performing standardized assistive tasks like bathing or grooming a mannequin. The robot occasionally exhibits scripted errors, such as missed grasps, early stops, or motion misalignment, while participants choose when and how to intervene. Alongside these physical tasks, participants complete concurrent mock web-based caregiving tasks to simulate real-world multitasking.

We collect both quantitative and qualitative measures of trust, perceived capability, willingness to intervene, and future adoption preferences, comparing results across caregiver types. The goal is to inform the design of trust-aware, adaptive robotic systems that respect the diverse roles and expectations of caregivers.

This work is conducted under Dr. Vaibhav Unhelkar in the Human-Centered AI and Robotics Group at Rice University.