Impact-Angle and Terminal-Maneuvering-Acceleration Constrained Guidance against Maneuvering Target

A new, highly constrained guidance law is proposed against a maneuvering target while satisfying both impact angle and terminal acceleration constraints. Here, the impact angle constraint is addressed by solving an optimal guidance problem in which the target’s maneuvering acceleration is time-varying. To deal with the terminal acceleration constraint, the closed-form solutions of the new guidance are needed. Thus, a novel engagement system based on the guidance considering the target maneuvers is put forward by choosing two angles associated with the relative velocity vector and line of sight (LOS) as the state variables, and then the system is linearized using small angle assumptions, which yields a special linear time-varying (LTV) system that can be solved analytically by the spectral-decomposition-based method. For the general case where the closing speed, which is the speed of approach of the missile and target, is allowed to change with time arbitrarily, the solutions obtained are semi-analytical. In particular, when the closing speed changes linearly with time, the completely closed-form solutions are derived successfully. By analyzing the generalized solutions, the stability domain of the guidance coefficients is obtained, in which the maneuvering acceleration of the missile can converge to zero finally. Here, the key to investigating the stability domain is to find the limits of some complicated integral terms of the generalized solutions by skillfully using the squeeze theorem. The advantages of the proposed guidance are demonstrated by conducting trajectory simulations.

Missile guidance with assisted deep reinforcement learning for head-on interception of maneuvering target

In missile guidance, pursuit performance is seriously degraded due to the uncertainty and randomness in target maneuverability, detection delay, and environmental noise. In many methods, accurately estimating the acceleration of the target or the time-to-go is needed to intercept the maneuvering target, which is hard in an environment with uncertainty. In this paper, we propose an assisted deep reinforcement learning (ARL) algorithm to optimize the neural network-based missile guidance controller for head-on interception. Based on the relative velocity, distance, and angle, ARL can control the missile to intercept the maneuvering target and achieve large terminal intercept angle. To reduce the influence of environmental uncertainty, ARL predicts the target’s acceleration as an auxiliary supervised task. The supervised learning task improves the ability of the agent to extract information from observations. To exploit the agent’s good trajectories, ARL presents the Gaussian self-imitation learning to make the mean of action distribution approach the agent’s good actions. Compared with vanilla self-imitation learning, Gaussian self-imitation learning improves the exploration in continuous control. Simulation results validate that ARL outperforms traditional methods and proximal policy optimization algorithm with higher hit rate and larger terminal intercept angle in the simulation environment with noise, delay, and maneuverable target.