Formation path planning based on quantum genetic algorithm for AUV

Abstract To address the challenges of significant spacing fluctuations and large tracking errors in autonomous underwater vehicle formations operating in complex obstacle-rich environments, this paper proposes a path planning method based on an improved quantum genetic algorithm (QGA). Firstly, to overcome the inherent limitations of traditional GAs, such as path redundancy and slow convergence, a QGA is proposed. It employs quantum bit encoding and superposition principles to enable parallel multi-path exploration, and integrates elite preservation, quantum tournament selection, and a dynamic mutation rate control mechanism to accelerate convergence. Secondly, to address the inherent drawbacks of the traditional artificial potential field (APF) method, including local minima, oscillatory motion, and abrupt heading changes, an improved APF (IAPF) method is introduced. This approach integrates a velocity-aligned adaptive attraction, an obstacle density-dependent repulsion, a tangential auxiliary force, and a maximum steering angle constraint, with spherical linear interpolation used to regulate the resultant force direction. Finally, comprehensive simulation experiments are conducted, including parameter sensitivity analysis, dynamic obstacle real-time avoidance tests, comparative experiments with particle swarm optimization (PSO) and rapidly-exploring random tree (RRT) algorithms, and dual-algorithm improvement validation. Simulation results demonstrate that the dynamic obstacle avoidance experiment verifies excellent real-time performance; the parameter sensitivity experiment confirms strong robustness to core parameter variations; and the comparison with PSO and RRT highlights significant superiority in path quality. The proposed QGA-IAPF framework fully leverages the complementary strengths of both algorithms, enhancing formation adaptability in complex obstacle-rich environments, accelerating initial formation convergence, and reducing spacing variations in high-density obstacle regions. It effectively lowers collision risk, thereby improving operational safety and reliability.