A Novel Robot System Integrating Biological and Mechanical Intelligence Based on Dissociated Neural...

Area of Science:

• Neuroscience
• Robotics
• Artificial Intelligence

Background:

• Merging biological and artificial intelligence presents challenges in creating adaptive robotic systems.
• Neural networks offer potential for advanced control but require effective integration with artificial agents.

Purpose of the Study:

• To develop and test a novel robot system architecture integrating a biological neural controller with an artificial agent.
• To investigate the use of a hierarchical dissociated neural network for controlling a mobile robot.
• To explore the impact of neural plasticity on robotic task performance.

Main Methods:

• A framework connecting a dissociated neural network to a mobile robot with a camera and two-wheeled drive was established.
• A custom stimulation generator facilitated bi-directional communication between biological neural cultures and the artificial robot via binomial coding.
• A hierarchical dissociated neural network was employed as the neural controller for a target-searching task.

Main Results:

• Neural cultures successfully controlled the artificial robot, achieving high performance in target searching.
• Robot performance improved with increased training cycles, demonstrating the effect of short-term plasticity (reinforced learning) in the neural network.
• The study preliminarily showed that neural network plasticity development independently improved robot performance.

Conclusions:

• This novel framework successfully demonstrates the control of an artificial agent by neural cultures, paving the way for intelligent robots.
• The findings suggest that engineering neural network plasticity offers solutions for enhancing robot learning abilities.
• The research provides theoretical inspiration for next-generation neuro-prosthetics based on bi-directional information exchange within hierarchical neural networks.