An Improved Genetic Algorithm for Path-Planning of Unmanned Surface Vehicle

Area of Science:

• Robotics and Autonomous Systems
• Artificial Intelligence
• Marine Engineering

Background:

• Genetic algorithms (GA) are widely used for path-planning in autonomous navigation.
• Conventional GAs face challenges like premature convergence and slow speed.
• Unmanned Surface Vehicles (USVs) require efficient navigation and control systems.

Purpose of the Study:

• To improve the conventional genetic algorithm (GA) for enhanced path-planning.
• To address limitations of traditional GAs, specifically premature population convergence and slow convergence rates.
• To optimize the navigation, guidance, and control of Unmanned Surface Vehicles (USVs).

Main Methods:

• Proposed a novel strategy involving multi-domain inversion to increase offspring generation.
• Implemented a secondary fitness evaluation to filter and retain superior individuals.
• Conducted Monte-Carlo simulations using Traveling Salesman Problem (TSP) datasets.
• Applied the improved algorithm to a USV's navigation system in a real maritime setting.

Main Results:

• The enhanced GA demonstrated improved local search capabilities and a higher probability of generating optimal solutions.
• Simulations showed the algorithm's effectiveness on TSP examples.
• Application to USV navigation confirmed superior performance compared to conventional methods.
• The multi-domain inversion approach resulted in a shorter optimal path and faster convergence speed.

Conclusions:

• The proposed multi-domain inversion strategy significantly enhances GA performance for path-planning.
• The improved algorithm offers a better balance between path length and time-cost for USV navigation.
• The enhanced GA exhibits superior robustness and efficiency in real-world maritime environments.