Target Acquisition for Projectile Vision-Based Navigation.

摘要

The application of computer vision to projectile navigation is a novel and untested proposition. As such, it is natural to wonder at this early stage how optical constraints may impact system performance. In this report, we develop a model for visual acquisition for a strap-down forward-facing imager with a constant field of view. Several prototypical scenarios are modeled, including stationary and moving target engagements. We further address the question of where to optimally place the initial ballistic impact point in relation to targets of uncertain position and heading. Finally, results of Monte Carlo simulations are presented to characterize the effects of target location error on visual acquisition.