Direct Adaptive Guidance - A NeuralNetworks Based Implementation

摘要

The predictive guidance problem of a vehicle so to minimize control eort isconsidered. The goal of the suggested approach is to reduce the computation loadrequirements for onboard calculations. It combines a neural network to calculate thezero-eort-miss and time-to-go of the trajectory with a direct adaptive control lawwhich improves the sensitivity to measurement and calculation noise. The guidancescheme is demonstrated using a three-degrees of freedom simulation of a guidedvehicle.