Research on Automatic Trajectory Planning Method of Unmanned Ships Based on Multi-Objective Optimization

Abstract

For unmanned ships’ autonomous navigation and water traffic systems, trajectory planning is a critical problem. Because conventional optimization techniques only consider a single target, such as the shortest trajectory distance or lowest time, multi-objective optimization can generate trajectory solutions that are optimized in terms of both distance and smoothness, thereby better serving the control of unmanned vessels in real-world applications. To achieve this, we propose a multi-objective non-linear trajectory planning method based on an improved genetic algorithm. This method simultaneously considers both the trajectory distance and smoothness, while incorporating constraints related to the safety characteristics and inherent properties of the unmanned ship, including speed and steering angles. According to simulation data, the proposed multi-objective improved genetic algorithm in this paper is compared to the single-objective genetic algorithm that only considers path length in various environments, resulting in an average reduction of 18.4% in the number of turning points and 26.38% in average turning angles in the planned paths. Compared to the traditional $\text{A}\ast $ algorithm and PSO algorithm, it achieved an average reduction of 49.9% in the number of turning points and 78.48% in average turning angles. In practical cases, it demonstrated an average reduction of 30.47% in the number of turning points and 35.49% in average turning angles compared to other algorithms.

Keywords

• Unmanned ship control
• trajectory planning
• multi-objective optimization
• genetic algorithm