Robot Navigation Based on Potential Field and Gradient Obtained by Bilinear Interpolation and a Grid-Based Search

Area of Science:

• Robotics
• Artificial Intelligence
• Computer Science

Background:

• The artificial potential field method is a common approach for robot path planning.
• A significant limitation of this method is the occurrence of local minima, which can trap robots.
• Existing extensions aim to overcome this weakness, but often with trade-offs in complexity or performance.

Purpose of the Study:

• To propose a smooth navigation function that overcomes the local minima problem in robot path planning.
• To develop a method that generates smooth paths with moderate computational complexity.
• To ensure adaptability to dynamic environmental changes.

Main Methods:

• A smooth navigation function is proposed, integrating Dijkstra-based discrete static potential field evaluation with modified bilinear interpolation.
• The A* algorithm is employed to handle dynamic environmental changes by bypassing the static plan.
• Modifications to bilinear interpolation are developed for path-planning applicability.

Main Results:

• The proposed method effectively solves the local minima problem inherent in artificial potential fields.
• Smooth paths are generated with a balance between computational complexity and preservation of the static plan.
• The integration with the A* algorithm allows for efficient adaptation to dynamic environmental changes.

Conclusions:

• The developed navigation model offers a robust solution for robot path planning, addressing critical limitations of existing methods.
• It provides a computationally efficient approach to generating smooth, optimal paths in static and dynamic environments.
• The strategy demonstrates significant advantages in various test environments, enhancing robot navigation capabilities.