Closed Loop Interactions between Spiking Neural Network and Robotic Simulators Based on MUSIC and ROS

Area of Science:

• Computational Neuroscience
• Robotics
• Artificial Intelligence

Background:

• Assessing simulated neural systems requires realistic stimuli.
• Creating reproducible, manipulable, and closed-loop stimuli is technically challenging.

Purpose of the Study:

• To present a novel approach for coupling robotics and neural network simulators.
• To enable efficient, real-time closed-loop experiments for neurorobotics and computational neuroscience.

Main Methods:

• Implemented middleware connecting Robotic Operating System (ROS) to Multi-Simulator Coordinator (MUSIC).
• Enabled coupling of robotic and neural simulators via standardized interfaces.

Main Results:

• Achieved efficient, real-time performance for coupled robotic and neural simulations.
• Facilitated a wide range of experimental configurations.

Conclusions:

• The developed system extends research tools for neurorobotics and computational neuroscience.
• Opens opportunities for complex closed-loop experiments on embodiment, agency, and reinforcement learning.