Vision-based roll and pitch attitude and rate estimation algorithms were developed for gun-launched precision projectiles. The gun-launched environment features unique challenges to interpreting data from a strapped-down imager on a projectile experiencing roll rates in the range of tens or even hundreds of revolutions per second. The goal of the present work is to develop vision-based algorithms for estimating the attitude of projectiles. The roll and pitch attitudes are determined using two horizon detection methods, the Hough transform and intensity standard deviation. The Hough transform method determined roll angle to better than 0.1° error variance in experiments when a forward-looking imager was spun. The intensity standard deviation algorithm estimated attitude to an average of 1° and 3° and standard deviation of 4° pitch and 6° roll in experiments with a side-facing imager.
展开▼