A Path-Planning Approach Based on Potential and Dynamic Q-Learning for Mobile Robots in Unknown Environment.

Area of Science:

• Robotics
• Artificial Intelligence
• Machine Learning

Background:

• Path planning is crucial for mobile robot navigation efficiency.
• Unknown environments present significant challenges for traditional path-planning algorithms.
• Existing methods often struggle with computational time and convergence speed.

Purpose of the Study:

• To propose a novel path-planning algorithm for mobile robots in unknown environments.
• To enhance the efficiency and practicality of mobile robot navigation.
• To improve upon the limitations of classical path-planning techniques.

Main Methods:

• A potential and dynamic Q-learning (PDQL) approach was developed.
• The PDQL algorithm integrates Q-learning with artificial potential fields and dynamic reward functions.
• Path feasibility is generated through the combined algorithmic approach.

Main Results:

• The PDQL approach demonstrated significant improvements in computing time and convergence speed.
• Experiments showed superior performance compared to state-of-the-art algorithms in simulated environments.
• Key metrics, including path length and turning angle, were optimized by the PDQL method.

Conclusions:

• The PDQL algorithm is effective and practical for mobile robot path planning in unknown environments.
• The proposed method offers a substantial advancement over classical Q-learning counterparts.
• PDQL provides a robust solution for optimizing navigation efficiency in complex, unexplored terrains.