A biologically inspired meta-control navigation system for the Psikharpax rat robot

Area of Science:

• Robotics
• Computational Neuroscience
• Artificial Intelligence

Background:

• Previous simulations validated parts of the navigation model for rat-like behavior in mazes.
• The model's capacity to function on a real robot platform remained untested.

Purpose of the Study:

• To implement and test a biologically inspired navigation system on the Psikharpax robot.
• To evaluate the robot's self-localization and autonomous navigation capabilities in unknown environments.
• To investigate adaptive strategy selection inspired by rat prefrontal cortex functions.

Main Methods:

• Implementation of two navigation strategies: place-based planning and cue-guided taxon.
• Development of a strategy selection meta-controller using reinforcement learning.
• Integration of a context detector for environmental adaptation and strategy restoration.

Main Results:

• The Psikharpax robot successfully navigated unknown environments using implemented strategies.
• The robot learned to memorize and select optimal strategies based on environmental context.
• The system demonstrated adaptive behavior, mirroring rat performance in strategy shifting.

Conclusions:

• The biologically inspired navigation system is effective on a real robot platform.
• The meta-controller provides adaptive capabilities, advancing robotic learning architectures.
• The work offers computational insights into the rat prefrontal cortex's role in strategy shifting.