This paper proposes an adaptive multi-population particle swarm optimization (AMP-PSO) algorithm for collaborative path planning of multiple autonomous underwater vehicles (AUVs). The algorithm introduces a novel grouping strategy and dynamic particle exchange mechanism to enhance global search capability, convergence speed, and solution adaptability in complex underwater environments. AMP-PSO separates particles into a leader population and various follower populations based on their fitness. The leader population learns from follower populations to improve global optimization, while follower populations focus on local exploration. An adaptive parameter configuration is also included to enhance the algorithm's performance. The algorithm is tested in various scenarios of collaborative path planning for multiple AUVs in an underwater environment. Simulation results show that AMP-PSO outperforms classic PSO and other improved PSO methods in terms of path planning efficiency and solution quality. The algorithm effectively avoids collisions and meets navigation constraints, enabling multiple AUVs to achieve their objectives. The main contributions of this study include the development of a distributed multi-population strategy for collaborative path planning and an adaptive parameter configuration to enhance multi-population performance. The results demonstrate that AMP-PSO is a significant advancement in efficient and effective AUV path planning.This paper proposes an adaptive multi-population particle swarm optimization (AMP-PSO) algorithm for collaborative path planning of multiple autonomous underwater vehicles (AUVs). The algorithm introduces a novel grouping strategy and dynamic particle exchange mechanism to enhance global search capability, convergence speed, and solution adaptability in complex underwater environments. AMP-PSO separates particles into a leader population and various follower populations based on their fitness. The leader population learns from follower populations to improve global optimization, while follower populations focus on local exploration. An adaptive parameter configuration is also included to enhance the algorithm's performance. The algorithm is tested in various scenarios of collaborative path planning for multiple AUVs in an underwater environment. Simulation results show that AMP-PSO outperforms classic PSO and other improved PSO methods in terms of path planning efficiency and solution quality. The algorithm effectively avoids collisions and meets navigation constraints, enabling multiple AUVs to achieve their objectives. The main contributions of this study include the development of a distributed multi-population strategy for collaborative path planning and an adaptive parameter configuration to enhance multi-population performance. The results demonstrate that AMP-PSO is a significant advancement in efficient and effective AUV path planning.