Pupillometry for Arm and Hand Motor Intent Detection.

Area of Science:

• Rehabilitation Robotics
• Human-Computer Interaction
• Neuroscience

Background:

• Assistive devices and rehabilitation robots benefit from intuitive user control.
• Detecting motor intent is crucial for effective human-robot interaction.
• Pupil dilation is a physiological response to cognitive and motor activity.

Purpose of the Study:

• To investigate the feasibility of using pupillometric data to differentiate between the motor intents of picking up versus observing an object.
• To explore the potential of pupil dilation as a non-invasive BCI (Brain-Computer Interface) signal for intent detection.

Main Methods:

• A human participant study involving 30 individuals performing object interaction tasks.
• Recording pupil dilation using an eye-tracking headset during motor tasks.
• Utilizing time-series feature extraction and a neural network classifier with leave-one-out cross-validation.

Main Results:

• The neural network classifier achieved an average accuracy of 59.4% and an F1 score of 0.578.
• Significant inter-participant variability in classifier performance was observed.
• The findings suggest that pupillometry-based intent detection is participant-dependent.

Conclusions:

• Pupil dilation shows limited but potential utility for detecting motor intent in human-robot interaction.
• Further research is needed to optimize machine learning models (e.g., CNNs) and explore real-time applications.
• Individual differences in pupillometric responses necessitate personalized approaches for effective intent detection.