Toward Perceiving Robots as Humans: Three Handshake Models Face the Turing-Like Handshake Test

Area of Science:

• Robotics
• Human-Computer Interaction
• Artificial Intelligence

Background:

• The Turing test assesses machine intelligence by indistinguishable human-like responses.
• Robotic movement generation lacks a standardized test for human-likeness.

Purpose of the Study:

• To develop a Turing-like test for evaluating the human-likeness of robotic handshakes.
• To compare the human-likeness of different robotic handshake models.

Main Methods:

• A telerobotic system was used to administer a handshake test.
• An interrogator interacted with artificial or human parties via a robotic stylus.
• Participants judged which interaction felt more human-like.

Main Results:

• The Tit-for-Tat and Machine Learning models generated the most human-like handshakes.
• These models outperformed the λ model in perceived human-likeness.

Conclusions:

• The developed Turing-like handshake test effectively evaluates robotic movement human-likeness.
• Combining elements from the best-performing models could enhance future robotic handshake algorithms.
• This research contributes to understanding sensorimotor control and improving human-robot interaction.