.png){kind=icon}
.png){kind=icon}
Return
Graph-reinforcement-learning-based distributed path planning for collaborative multi-AGV systems
.png){kind=icon}
PRE
AI
DOI：10.1016/j.knosys.2025.114255.png)
.png){kind=icon}
Original request for help.png){kind=icon}
Share.png){kind=icon}
Save
Abstract
En 中文
• A Graph-Reinforcement-Learning-based Path Planning (GNMAPP) algorithm is proposed for information-constrained multi-AGV systems. • GNMAPP integrates dynamic heterogeneous graphs and GNN for enhanced AGV-environment interaction. • The algorithm combines a multi-head attention mechanism and RRT-guided reward to improve exploration. • Experimental results show that GNMAPP outperforms existing algorithms in decision-making efficiency and convergence speed.
Keywords:
Graph-Reinforcement Learning
Multi-AGV Systems
Path Planning
Dynamic Heterogeneous Graphs
GNN