Bio-robots automatic navigation with graded electric reward stimulation based on Reinforcement Learning

Area of Science:

• Robotics
• Neuroscience
• Artificial Intelligence

Background:

• Bio-robots utilizing brain-computer interfaces (BCI) often neglect inherent animal navigation characteristics.
• Effective bio-robot navigation requires integrating animal intelligence with algorithmic approaches.

Purpose of the Study:

• To develop a novel method for bio-robot automatic navigation by combining Reinforcement Learning (RL) with animal learning intelligence.
• To improve bio-robot navigation efficiency by leveraging animals' spatial recognition and reward-seeking behavior.

Main Methods:

• Proposed a reward-generating algorithm based on Reinforcement Learning (RL).
• Integrated RL with the learning intelligence of animals, using rats as a model.
• Employed graded electrical rewards to guide animal exploration and learning.

Main Results:

• The rat-robot and RL algorithm achieved convergence to an optimal navigation route through co-learning.
• Demonstrated that animals' spatial recognition significantly aids in optimizing navigation paths.
• Validated the effectiveness of hybrid intelligence in bio-robot navigation.

Conclusions:

• The proposed hybrid intelligence method offers a significant advancement for bio-robot navigation systems.
• Integrating animal learning with RL provides a promising direction for developing more intelligent and adaptive bio-robots.
• This approach inspires practical applications in bio-robotics, enhancing navigation capabilities.