Generalized Reinforcement Learning Enhanced Three-Dimensional Impact Time Control Against Maneuvering Targets With Field-of-View Constraint

This paper presents a generalized three-dimensional (3D) intelligent impact time control guidance (ITCG) law for intercepting maneuvering targets under the field-of-view (FOV) constraint in the relative virtual framework (RVF). The proposed method addresses key limitations of existing guidance approaches by eliminating the need for time-to-go estimation, model linearization, and decoupling. First, a generalized ITCG law with the FOV constraint is derived by employing two newly defined impact errors and the backstepping control technique. Second, a future average speed prediction algorithm is subsequently introduced. Third, the proximal policy optimization (PPO) algorithm is employed to optimize the controller gain (Formula presented.), which minimizes both guidance errors and energy consumption. Finally, a novel implementation case further illustrates the key advantage of the proposed ITCG law, whose impact time feasible region can be derived analytically. Unlike other existing guidance laws, the proposed approach does not rely on restrictive assumptions or approximations, which makes it more applicable for actual combat scenarios. Numerical simulations validate the superiority of the proposed ITCG law over existing methods in terms of accuracy, robustness, and energy efficiency.