Spatiotemporal Modeling of Grip Forces Captures Proficiency in Manual Robot Control.

Area of Science:

• Systems Neuroscience
• Cognitive Behavioral Science
• Human-Robot Interaction

Background:

• Somatosensory cognition is crucial for object manipulation and force control.
• Previous work utilized wireless grip-force tracking for insights into perceptual and motor learning with unreliable sensory input.

Purpose of the Study:

• To analyze grip-force data in manual robot control using functionally motivated approaches.
• To develop and apply a brain-inspired neural network model for assessing skill levels.

Main Methods:

• Recorded grip forces using wearable wireless sensor technology from dominant and non-dominant hands.
• Applied statistical analyses and a self-organizing neural network with unsupervised learning to sensor data.
• Conducted spatiotemporal analysis of grip forces from task-relevant sensors in the dominant hand.

Main Results:

• Identified skill-specific temporal variations in grip forces differentiating novices and experts.
• The neural network metric reliably captured differences in grip-force variability between skill levels.
• Revealed finger-specific synergies reflecting robotic task skill in dominant hand grip forces.

Conclusions:

• Grip-force monitoring can track skill evolution and proficiency in human-robot interaction.
• This technology has potential applications in improving surgical outcomes, particularly in single-port surgery.
• Cross-disciplinary insights from neuroscience, behavioral science, and AI are key for predictive operator skill modeling.