The article introduces OkayPlan, a Global Path Planning (GPP) algorithm that allows Unmanned Surface Vehicles (USVs) to navigate efficiently in dynamic environments. It overcomes the limitations of existing GPP algorithms that mostly assume static environments. Using Particle Swarm Optimization, OkayPlan can generate safe and short paths in dynamic scenarios at a real-time speed. It also includes a Dynamic Prioritized Initialization mechanism for improving solution quality and a relaxation strategy for autonomous tuning of hyperparameters in dynamic environments. Experimental results demonstrate that OkayPlan outperforms existing methods in terms of path safety, length optimality, and computational efficiency.
Publication date: 12 Jan 2024
Project Page: https://github.com/XinJingHao/OkayPlan
Paper: https://arxiv.org/pdf/2401.05019
