Learning robotic eye-arm-hand coordination from human demonstration: a coupled dynamical systems approach.

Area of Science:

• Neuroscience
• Robotics
• Human Motor Control

Background:

• Visually guided reaching and grasping are fundamental motor skills.
• Obstacle avoidance is a critical component of naturalistic movement.
• Understanding the sensorimotor control underlying these actions is key for human-robot interaction.

Purpose of the Study:

• To investigate the role of obstacle avoidance in visually guided reaching and grasping.
• To analyze the sequential organization and planning involved in obstacle avoidance.
• To quantify the eye-arm-hand coordination during obstacle avoidance and develop a computational model.

Main Methods:

• Human participants performed prehensile movements with systematically varied obstacle positions.
• Analysis of movement trajectories, gaze patterns, and eye-arm-hand coupling.
• Development and validation of a computational model based on coupled dynamical systems.

Main Results:

• Reaching with obstacle avoidance is organized sequentially, with obstacles acting as intermediate targets.
• A forward planning scheme actively detects obstacles during reaching.
• Gaze proactively coordinates eye-arm motion, and eye-arm-hand coupling remains consistent during obstacle avoidance.

Conclusions:

• Obstacle avoidance is an integrated process within reaching and grasping, involving predictive planning and coordinated sensorimotor control.
• The developed computational model effectively mimics human visuomotor control for obstacle avoidance.
• The findings have implications for designing advanced humanoid robots capable of navigating complex environments.