Desired Impact Angle Identification for An Incoming Aerial Vehicle Using the Trajectory Shaping Guidance Law

This paper investigates the problem of impact angle identification for an incoming aerial vehicle, which is essential for constructing an effective Kalman Filter (KF) and accurately estimating the vehicles state. The considered scenario is that a vehicle attempts to hit a stationary target with the Trajectory Shaping Guidance (TSG) law. A desired impact angle regression identification model from the perspective of the target is proposed based on a Gated Recurrent Unit (GRU) neural network. The input of the model is a period of available information, including the relative distance and relative angle between the vehicle and target, while the output is the regression identification result. To increase the offline training efficiency and improve the online identification accuracy, the Multiple-Model Mechanism (MMM) is introduced into the network. Simulation results demonstrate the advantage of the proposed model over a conventional network, and verify its application potential.