Human Perception in Social Tasks: A Comparative Evaluation of Autonomous and Teleoperated Robots
Lavinia Hriscu, Alberto Sanfeliu, Anais Garrell
AI summary
Problem
How do users perceive and evaluate robots differently when controlled by a human versus an autonomous AI system during real-time social interactions?
Approach
We conducted a controlled user study where participants interacted with a robot performing information and delivery tasks under both autonomous (LLM) and teleoperated conditions, measuring perceived intelligence, safety, and satisfaction across static and dynamic movement modes.
Key results
- Human operators outperformed autonomous systems in spatial and contextual tasks
- Autonomous LLM robots were rated more favorably for rapid information access
- Robot mobility (static vs. dynamic) did not significantly affect perceived intelligence
- Repeated interactions slightly increased perceived intelligence ratings
Why it matters
Results help designers and researchers choose between autonomous AI and human teleoperation to optimize user trust, acceptance, and task performance in social robotics.
Abstract
Robots and artificial intelligence technologies are be- coming increasingly integrated into our daily lives. The introduc- tion of humanoid robots into everyday settings is a gradual but ongoing process—one that society is already beginning to navigate. Yet this shift raises important questions: Who or what is truly behind these physical agents? And can we, as users, perceive dif- ferences in our interactions depending on whether a robot acts autonomously or it’s teleoperated by a human? In this study, we present the results of an experiment in which participants interacted with a robot under two control conditions—autonomous and teleoperated—while it performed two distinct tasks in both static and dynamic movement scenarios. In our results, human operators outperformed autonomous systems in tasks requiring spatial awareness and contextual reasoning. Conversely, the au- tonomous robot—powered by a Large Language Model and oper- ating without visual input—was perceived more favorably in tasks that demanded rapid access to broad and diverse information.