Robot path planning algorithm with improved DDPG algorithm

• Published in: International journal on interactive design and manufacturing Vol. 19; no. 2; pp. 1123 - 1133
• Main Author: Lyu, Pingli
• Format: Journal Article
• Language: English
• Published: Paris Springer Paris 01.02.2025; Springer Nature B.V
• Subjects: Algorithms; Autonomous navigation; CAE) and Design; Computer-Aided Engineering (CAD; Deep learning; Electronics and Microelectronics; Engineering; Engineering Design; Industrial Design; Instrumentation; Machine learning; Mechanical Engineering; Neural networks; Obstacle avoidance; Original Article; Path planning; Planning; Robot control; Robotics; Robots; Sensors; China; DDPG; Deep reinforcement learning; Path planning; Artificial potential field; Mobile robot
• ISSN: 1955-2513; 1955-2505
• DOI: 10.1007/s12008-024-01834-x

This study focuses on enhancing the autonomous path planning capabilities of intelligent mobile robots, which are complex mechatronic systems combining various functionalities such as autonomous planning, behavior control, and environment sensing. Path planning is crucial for robot mobility, enabling them to navigate autonomously. We propose an improvement to the deep deterministic policy gradient (DDPG) method by leveraging deep reinforcement learning algorithms. Through extensive experimentation, our method demonstrates superior performance compared to traditional DDPG, with notable reductions in training time and iterations required to reach targets. Additionally, it reduces dead zone encounters during travel and enhances convergence speed. Our findings contribute fresh insights and strategies for enhancing mobile robot path planning in unfamiliar environments. Future research will explore further advancements, particularly in addressing dynamic obstacles and optimizing real-world navigation efficiency.