Adaptive-dynamic-programming-based cooperative active defense guidance against a guided attacker

This paper develops an adaptive-dynamic-programming-based cooperative defense guidance law to enable a defender to protect a target against a superior maneuvering attacker. The cooperative defense is firstly formulated as a line-of-sight tracking problem with the cost-optimal error dynamics. To eliminate the tracking error, a concurrent bellman error extrapolation learning method is proposed to integrate into the critic neural network (NN) and the actor NN, which iteratively updates the weights of the NNs, so as to approximate the optimal solution of the tracking problem online. Then, a two-way cooperation strategy that minimizes the combined steady-state control effort is proposed to reduce the lateral acceleration demand of the defender. Finally, an identifier is developed for the attacker employing a structurally unified class of guidance laws, using singular value maximization-based least squares technique. The superiority of the proposed cooperative defense guidance law is demonstrated through simulations in different engagement scenarios.