Multi-AUV path planning based on a multi-strategy fused improved nutcracker optimization algorithm

This is the authors' abstract. We don't add key points for this paper.

Autonomous underwater vehicles (AUVs) have emerged as pivotal platforms for deep-sea exploration and offensive and defensive operations. Achieving path optimization under constraints, including path length and obstacle-avoidance safety, remains a challenge. The Nutcracker Optimization Algorithm (NOA) shows prospects for solving optimization problems, yet has low search accuracy and a tendency to be trapped in local optima in multi-AUV path planning in complex environments. To address these limitations, an improved Nutcracker Optimization Algorithm (INOA-SQP) is proposed, with a hybrid initialization method to enhance quality, a dynamic hierarchical acceptance criterion for position updating, an elitist sequential quadratic programming algorithm to improve local search accuracy, and a chaotic dynamic lens-imaging opposition-based learning strategy to escape local optima in the later optimization stage. Results on the CEC2017 suite indicate that INOA-SQP achieves the best convergence accuracy on 75% of test functions. INOA-SQP is compared with six swarm intelligence algorithms, and its practicality is validated on two seabed datasets. Compared with the best-performing baseline, INOA-SQP reduces path length, collision threat, smoothness, and fitness by 38.56%, 24.72%, 74.89%, and 56.53%, respectively. These results demonstrate that INOA-SQP delivers excellent performance for path planning in three-dimensional marine environments, with superior robustness and search capability over swarm intelligence algorithms.