In many applications, the automated unmanned underwater vehicles (UUVs) equipped with different kinds of sensors have been utilized to localize a beacon target in the sea. This paper is concerned with the target localization using one optimal path planned mobile UUV. An interaction multiple model extended Kalman filter (IMM-EKF) using different types of measurements is developed to estimate the target state. To improve the estimation performance, a gradient ascent path optimization algorithm is proposed to steer the mobile UUV. The presented path optimization strategy minimizes the estimation mean squared error (MSE) by maximizing the determinant of the Fisher information matrix (FIM). The properties and effectiveness of the proposed algorithms are discussed and verified with simulation examples.