Abstract:Aiming at the problems of unmanned underwater vehicle (AUV) with many motion constraints as well as the low efficiency of path optimization and slow convergence speed of traditional genetic algorithm, an improved genetic algorithm path planning method for AUV is proposed. The algorithm selects the raster method to construct the environment, and uses the path length, smoothness and dangerous area as the evaluation function. The genetic algorithm population initialization process is improved by introducing a grid of surrounding points to improve the convergence speed, and at the same time combining the idea of catastrophe to avoid the population falling into the local optimal solution. The algorithm designs the AUV smoothing process and deletion process according to the constraints of the maximum turning angle of the AUV, which avoids the sharp stop and turn of the AUV navigation. The simulation and on-lake test results show that the improved genetic algorithm reduces the path length by 11.4% compared with the traditional genetic algorithm, accelerates the convergence speed by 20.0% and the convergence path meets the constraints of AUV navigation.
