A physical model of sensorimotor interactions during locomotion

Area of Science:

• Robotics
• Biomechanical Engineering
• Computational Neuroscience

Background:

• Human walking is a complex motor control process involving intricate neuromuscular coordination.
• Previous robotic models have simplified the neuromuscular system, limiting insights into human locomotion.
• Understanding the interplay between neural control and body dynamics is crucial for replicating natural gait.

Purpose of the Study:

• To develop a neurorobotic model of a bipedal robot that accurately replicates human walking's neuromuscular architecture.
• To investigate the role of central pattern generators (CPGs) and reflex systems in generating and stabilizing gait.
• To compare the robot's joint trajectories with human walking data for validation.

Main Methods:

• Constructed a bipedal robot body based on human muscular architecture principles, using strap-based muscles and load sensors.
• Implemented a neural architecture featuring a central pattern generator (CPG) with a half-center oscillator and phase-modulated reflexes, simulated via a spiking neural network.
• Integrated the CPG, reflex system, and body dynamics to achieve autonomous walking.

Main Results:

• The robot's walking cycle became entrained to its body dynamics through the interaction of the reflex system, body dynamics, and CPG.
• The CPG significantly enhanced gait stability against perturbations compared to a purely reflexive system.
• Joint trajectories generated by the robot were compared to human walking data, showing congruence.

Conclusions:

• This neurorobotic model provides a comprehensive physical simulation of human walking's neuromuscular control.
• The research highlights the effectiveness of CPGs and reflex integration for stable and adaptable bipedal locomotion.
• This approach offers a valuable platform for advancing research into the neurophysiological underpinnings of human and animal walking.