Real-time biomimetic Central Pattern Generators in an FPGA for hybrid experiments

Area of Science:

• Neuroscience
• Biomimetic Engineering
• Computational Neuroscience

Background:

• Central Pattern Generators (CPGs) are crucial for rhythmic biological functions like locomotion and heartbeat.
• Previous CPG models, like the Hodgkin-Huxley (HH) model, are computationally intensive.
• There is a need for efficient, real-time CPG hardware for hybrid systems.

Purpose of the Study:

• To develop a low-resource, real-time biomimetic digital hardware system for CPGs.
• To implement a network of 240 CPGs using a simplified Izhikevich model and a novel synapse model.
• To enable hybrid experiments combining biological tissue and artificial neural networks.

Main Methods:

• Implemented a network of 240 CPGs on a Field Programmable Gate Array (FPGA).
• Utilized the Izhikevich neural model and an activity-dependent depression synapse model.
• Validated the hardware implementation against complex HH model simulations and rat spinal cord pharmacological data.

Main Results:

• Achieved real-time operation with a single computation core and minimal resource utilization.
• The implemented CPG network exhibited bursting activity behavior comparable to complex models.
• Successfully matched simulation results with pharmacological data, validating the system's biological relevance.

Conclusions:

• The developed digital CPG system is resource-efficient and operates in real-time.
• This hardware facilitates hybrid experiments and advances neuroprosthesis development.
• The CPG network has potential applications in mimicking animal locomotion and brain-computer interfaces.