Terminal Guidance Laws: A Comprehensive Survey from Classical Methods to Learning-Based Approaches

Abstract

This survey reviews terminal guidance laws for guided munitions within a single, unified framework. Starting from the classical proportional navigation family (True, Pure, and Augmented Proportional Navigation), the review proceeds through practical corrections such as gravity and drag compensation, the generalized vector explicit guidance law (GENEX) that additionally controls the terminal impact angle, sliding mode-based robust guidance, differential game guidance grounded in the Hamilton Jacobi-Isaacs equation, and finally reinforcement learning based guidance. For each method the underlying derivation logic, the command equations, and the design trade offs are presented with numbered equations. The methods are compared along the axes of information requirement, robustness, computational load, and theoretical guaran tee. A recurring theme is that classical laws and learning-based methods are connected through the common zero-effort-miss and dynamic-programming principles, which motivates the current trend toward hybrid designs that combine the analytic reliability of classical laws with the flexibility of learning-based methods.