Flexible Spiking CPGs for Online Manipulation During Hexapod Walking.

Area of Science:

• Neuroscience
• Robotics
• Computational Biology

Background:

• Neural control of locomotion involves complex interactions between network architecture and sensory input for gait adaptation.
• Understanding the dynamics of the sensorimotor loop is crucial for developing closed-loop control systems.
• Adaptation of gait requires dynamic modulation of neural signal parameters (amplitude, frequency, phase).

Purpose of the Study:

• To develop a spiking neural network capable of online adaptation of locomotion parameters.
• To investigate the role of network topology in maintaining stable gait without sensory feedback.
• To enable further research into sensory feedback and high-level control for gait adaptation.

Main Methods:

• Developed a biologically inspired spiking neural network with online parameter adaptation capabilities.
• Implemented the network on a hexapod robot to observe locomotion behavior.
• Analyzed the resulting gait patterns and stability.

Main Results:

• The network successfully generated a stable tripod gait on the hexapod robot.
• The implemented network demonstrated that afferent feedback is not essential for maintaining a stable gait.
• Results align with biological findings on deafferented locomotion, supporting the role of network topology.

Conclusions:

• A spiking neural network architecture alone can produce stable locomotion, highlighting the importance of network topology.
• This work provides a platform for studying sensory feedback and higher-level control in gait adaptation.
• The findings offer insights into the neural control of locomotion, with potential applications in understanding biological systems.