실시간 탄도 궤적 목표물 추적을 위한 GPU 기반 병렬적 Monte Carlo 최적화 기법

Abstract

This paper seeks to achieve real-time high-speed ballistic target tracking. Monte Carlo optimization can be considered to solve the non-linearity in motion and measurement models in high-speed targets, but applying it to real-time tracking systems is difficult because it requires a long computation time. This paper proposes a graphics processing unit (GPU)-based parallelization method to accelerate Monte Carlo optimization for real-time ballistic target tracking. The improved performance of the proposed method was tested and analyzed on awidelyused embedded system. Comparisons with existing Monte Carlo optimization on a central processing unit (CPU) revealed that the proposed method improved the real-time performance by greatly reducing the computation time.