Time-optimal global path planning and collision-avoidance local path planning for USVs in traffic separation scheme-implemented coastal waters

• Published in: ISA transactions
• Main Authors: Tao, Yihan, Du, Jialu
• Format: Journal Article
• Language: English
• Published: United States Elsevier Ltd 04.07.2025
• Subjects: Gaussian process regression; Irregular static obstacle; Nonlinear programming; Path planning; Traffic separation scheme; Unmanned surface vehicle; Gaussian process regression; Unmanned surface vehicle; Path planning; Nonlinear programming; Irregular static obstacle; Traffic separation scheme
• ISSN: 0019-0578; 1879-2022; 1879-2022
• DOI: 10.1016/j.isatra.2025.06.030

Under multiple constraints including unmanned surface vehicle (USV) dynamics, traffic separation scheme (TSS) requirements, navigable water boundaries, and safety thresholds for collision risks, time-optimal path planning and collision-avoidance (COLAV) path planning for USVs in TSS-implemented coastal waters remain challenging. To overcome this challenge, we innovatively develop a hierarchical Gaussian-process-based nonlinear programming (GPNLP) approach for the USV time-optimal global path planning and COLAV local path planning. We model irregular static obstacles using Gaussian process regression for the first time, such that navigable waters are more sufficiently utilized for path planning. A TSS compliance assessment function is created to output violation penalties for the TSS requirements that should be satisfied as far as practicable. Accordingly, we plan the time-optimal global path and the COLAV local path hierarchically by minimizing two integral objective functions (with respect to the TSS violation penalties) subject to the multiple constraints. Simulations and simulation comparison results demonstrate that both the planned USV time-optimal global path and COLAV local path under the proposed hierarchical GPNLP approach are USV dynamics compliant and TSS compliant. •A traffic separation scheme (TSS) compliance assessment function is created to output violation penalties for the TSS requirements.•Irregular static obstacles are more precisely modeled using Gaussian process regression.•Both a planned USV time-optimal global path and a collision-avoidance local path comply with USV dynamics and TSS.