.png){kind=icon}
Return
A multi-objective path planning method for ships based on constrained policy optimization
.png){kind=icon}
PRE
AI
DOI：10.1016/j.oceaneng.2024.120165.png)
.png){kind=icon}
Original request for help.png){kind=icon}
Share.png){kind=icon}
Save
Abstract
En 中文
With development of technologies in ships, path planning plays a crucial role in ensuring efficient and safe navigation. However, traditional single-objective path planning methods suffer from inefficiency, and are prone to getting trapped in local optima. As for the traditional multi-objective path planning methods, challenges may also be faced for conflicts among objectives. To address these challenges, this study formulated the path planning problem as a multi-objective optimization model. Constrained policy optimization (CPO) offers key advantages by effectively handling constraints, optimizing and balancing objectives, and ensuring safety. Thus, this study proposed a multi-objective path planning method based on CPO to explore pareto-optimal paths. Results from four environments with different types and distributions of obstacles demonstrated that CPO-based method outperforms NSGA-II-based one and MOEA/D-based one in both path length and energy consumption. CPO-based method achieves the shortest path (261.89 m) compared to NSGA-II-based one (269.13 m) and MOEA/D-based one (267.83 m), and has the lowest energy consumption (1.10 x 106 J), significantly lower than NSGA-II-based one (5.32 x 106 J) and MOEA/D-based one (6.15 x 106 J). The ability of CPO-based method to maintain optimal Pareto fronts and to adapt to various weight settings established it as an advantageous method in complex environments.
Keywords:
Multi-objective path planning
MASS
Safe reinforcement learning
Constrained policy optimization
Journal
IF:
5.5
Papers:
7.7K
Citations:
7.6W
Organization
Cited Papers
Exploring the application of ant colony optimization in path planning for Unmanned Surface Vehicles
.png){kind=icon}
PREAI
ShareSave
ShareSaveShareSave
ShareSaveShareSave
View more