Autonomous Underwater Vehicle Geomagnetic Perceiving Navigation Based on Policy Gradient

With the advancement of oceanic technology, there is a growing interest in the research of Autonomous Underwater Vehicle (AUV) for marine resource exploration and development. This article proposes a geomagnetic perceiving navigation method based on policy gradient for AUV without prior geomagnetic maps. This method leverages the intrinsic physical resource of the Earth by combining the search strategy of the gradient descent algorithm with geomagnetic perceiving navigation. By establishing motion equations and computing the gradients of the objective function, the heading angle of the AUV can be determined. During the navigation process of AUV, the objective function of various geomagnetic parameters will gradually decrease and converge to the target value, indicating the achievement of the autonomous navigation task for autonomous underwater vehicle. Simulation experiments demonstrate that this method enables the AUV to reach the desired target location in the absence of prior geomagnetic maps. A comparison with the hexagonal path planning algorithm verifies the effectiveness and superiority of the policy gradient based geomagnetic perceiving navigation method.