Predicting humor effectiveness of robots for human line cutting

Area of Science:

• Robotics
• Human-Robot Interaction
• Artificial Intelligence

Background:

• Security robots face challenges in preventing undesirable human behaviors like line-cutting.
• Existing methods for robot intervention are often ineffective or perceived negatively by humans.

Purpose of the Study:

• To develop and evaluate a system that uses humorous phrases to deter human line-cutting behavior.
• To create a machine learning model for predicting the effectiveness of humor in specific social situations.

Main Methods:

• Collected 500 humorous phrases via crowdsourcing and simulated 13,000 line-cutting scenarios.
• Developed a humor effectiveness predictor using machine learning, considering situational context and perceived discomfort.
• Constructed a system to select optimal humorous phrases for line-cutting intervention.

Main Results:

• Humor effectiveness prediction is significantly improved by accounting for situational context and phrase-induced discomfort.
• Humorous phrases selected by the proposed system demonstrated greater effectiveness in discouraging line-cutting compared to non-humorous phrases.
• Video experiments confirmed the practical efficacy of humor-based interventions in security robotics.

Conclusions:

• Humorous communication is a viable and effective strategy for security robots to manage human behavior.
• The developed humor effectiveness predictor and selection system offer a novel approach to human-robot social interaction.
• Future research can explore diverse applications of humor in robot-mediated social control and assistance.