Multi-robot systems are significant in decision-making capabilities and applications, but avoiding collisions during movement remains a critical challenge. Existing decentralized obstacle avoidance strategies, while low in computational cost, often fail to ensure safety effectively. To address this issue, this paper leverages graph neural networks (GNNs) and deep reinforcement learning (DRL) to aggregate high-dimensional features as inputs for reinforcement learning (RL) to generate paths. Additionally, it introduces safety constraints through an artificial potential field (APF) to optimize these trajectories. Additionally, a constrained nonlinear optimization method further refines the APF-adjusted paths, resulting in the development of the GNN-RL-APF-Lagrangian algorithm. By combining APF and nonlinear optimization techniques, experimental results demonstrate that this method significantly enhances the safety and obstacle avoidance capabilities of multi-robot systems in complex environments. The proposed GNN-RL-APF-Lagrangian algorithm achieves a 96.43% success rate in sparse obstacle environments and 89.77% in dense obstacle scenarios, representing improvements of 59% and 60%, respectively, over baseline GNN-RL approaches. The method maintains scalability up to 30 robots while preserving distributed execution properties.

In this article, UP3, a novel unsupervised learning‐based predictive path planning framework for mobile robot navigation in partially observable environments is proposed. An offline unsupervised learning method is used for planner training to avoid frequent environment interactions or expert demonstration. The network employs attention‐based path optimization to generate the predictive path. Herein, the robot's safety is considered as a hard constraint, and the control barrier function constraint is used to design the collision loss. Additionally, a deep constraint correction module is designed to correct the predictive path, which aims to make the path satisfy the constraints. Herein, the approach through simulation and real‐world experiments is tested to verify the function of each module. In the results, it is indicated that UP3 performs well on dense and maze scenarios and has good generalization.