SLP-Improved DDPG Path-Planning Algorithm for Mobile Robot in Large-Scale Dynamic Environment

Area of Science:

• Robotics
• Artificial Intelligence
• Computer Science

Background:

• Robotic navigation in large-scale environments with dynamic obstacles presents significant challenges.
• Traditional Deep Deterministic Policy Gradient (DDPG) algorithms struggle with stability and real-time performance in such complex scenarios.

Purpose of the Study:

• To enhance the stability and efficiency of DDPG algorithms for robot path planning.
• To achieve a superior balance between stability and real-time performance in dynamic environments.

Main Methods:

• Proposes an improved Deep Deterministic Policy Gradient (DDPG) path planning algorithm.
• Incorporates Sequential Linear Path Planning (SLP) to generate sub-goals.
• Utilizes DDPG to follow these sub-goals for navigation and obstacle avoidance.

Main Results:

• The SLP-enhanced DDPG algorithm demonstrates superior performance in large-scale dynamic environments.
• Achieved a 12.33% improvement in success rate compared to traditional DDPG.
• Showed a 29.67% improvement compared to the A*+DDPG algorithm.

Conclusions:

• The proposed SLP-enhanced DDPG algorithm effectively navigates robots in complex dynamic environments.
• Offers improved stability, efficiency, and success rates for robotic path planning.