Realistic 3D human saccades generated by a 6-DOF biomimetic robotic eye under optimal control

Area of Science:

• Biomimetics and Robotics
• Neuroscience and Computational Modeling

Background:

• The human eye's complex biomechanics and neural control of saccades are not fully understood.
• Previous models often simplify the nonlinear dynamics and multi-tendon system of the eye.

Purpose of the Study:

• To develop and analyze a biomimetic robotic eye system that accurately mimics human eye movement dynamics.
• To investigate optimal control principles underlying saccadic eye movements using a computational model.

Main Methods:

• Constructed a biomimetic robotic eye with six independently controlled tendons.
• Developed an accurate physical computational model and approximated it with a NARX recurrent neural network.
• Optimized motor control by minimizing saccade inaccuracy, duration, and energy expenditure.

Main Results:

• The robotic eye generated realistic saccades matching human main-sequence dynamics and Listing's law.
• Control signals exhibited agonist-antagonist muscle pairing and pulse-step characteristics similar to monkey motoneurons.
• The model successfully replicated nonlinear dynamics and cross-couplings in horizontal and vertical eye movements.

Conclusions:

• The biomimetic robotic eye system provides a valuable platform for studying oculomotor control.
• Optimal control principles may explain diverse behaviors in the primate oculomotor system.
• The model offers insights into neural control signals and their couplings in final neural pathways.