Efficient Path Planning for Maritime Search and Rescue: A Surrogate-Assisted Memetic Algorithm

Developing effective search paths to cover the movements of survivors or lost objects drifting in the ocean is critical for maritime rescue operations, which can save many lives. The goal of this study is to generate more efficient and realistic search paths while reducing execution time. A prior study used a memetic algorithm (MA) to develop search paths that consider particle drift movements, showing improved particle coverage compared to simple paths. However, the time required posed limitations for practical real-time applications. To overcome these limitations, we propose an improved approach that integrates surrogate models into the MA framework. We modify the objective function by incorporating probabilistic coverage, which reflects the uncertainty of locating drifting particles in real-world scenarios. Two encoding methods were applied and experiments were conducted on a 6×9 grid to evaluate the effects of model and encoding combinations. Among the combinations tested, the continued fraction regression model combined with one of the encoding methods achieved the best performance. Experimental results indicate that the proposed surrogate-assisted MA reduces execution time by 70% while maintaining a comparable performance. These findings suggest that the proposed approach has significant potential for real-time rescue operations.