Visual Perception-Based Target Aircraft Movement Prediction for Autonomous Air Combat

摘要

This paper introduces a framework that is constructed for target aircraft movement prediction for the autonomous air combat technique. This work is inspired by the common experience that human pilots could foresee future moves of the opponent in combat based on a visual check on the target. Movement of the target is depicted in body coordinates of own aircraft; an infrared/optical image of the target is adopted; orientation of target aircraft is estimated based on image moment-invariants analysis of neural networks; and the target future position is extrapolated based on current position and the estimated orientation. This paper also includes a test report of the proposed framework for classic combat maneuvers, which justifies applicability of the proposed framework.