CAJ | 학술논문

针对非协作式无人机检测与避障系统，采用多传感器进行信息融合的方式进行检测与跟踪，提出了采用正交积分点卡尔曼滤波(QKF)实时跟踪运动目标以提高检测精度和增强有效性。首先，对设计的检测与避障系统进行了简述，由两个子系统构成：由捷联惯性导航系统(SINS)与GPS组成的导航单元及由雷达和光电传感器组成的检测单元。其次，以拐弯模型与Singer模型两个机动运动模型为例测试了QKF算法跟踪检测障碍物的性能，并与无迹卡尔曼滤波(UKF）进行比较。仿真结果表明，相比于UKF算法，QKF算法可以更快速、更准确的检测与跟踪目标。
침대비협작식무인궤검측여피장계통，채용다전감기진행신식융합적방식진행검측여근종，제출료채용정교적분점잡이만려파(QKF)실시근종운동목표이제고검측정도화증강유효성。수선，대설계적검측여피장계통진행료간술，유량개자계통구성：유첩련관성도항계통(SINS)여GPS조성적도항단원급유뢰체화광전전감기조성적검측단원。기차，이괴만모형여Singer모형량개궤동운동모형위례측시료QKF산법근종검측장애물적성능，병여무적잡이만려파(UKF）진행비교。방진결과표명，상비우UKF산법，QKF산법가이경쾌속、경준학적검측여근종목표。
An improved algorithm utilizing quadrature Kalman filter(QKF) is proposed to locate a movable obstacle in real time for a “see and avoid” (S&A) system with multi-sensor detection in a non-cooperative unmanned aircraft vehicle(UAV) with enhanced accuracy and effectiveness. The S&A system primarily include two subsystems: SINS/GPS navigation unit and detection unit encompassing radar and EO sensors. The QKF algorithm fusing heterogeneous sensing data is exemplified by testing two dynamic nonlinear motion models of the obstacle, i.e. the turning model and the Singer model. In comparison with the results obtained from unscented Kalman filter(UKF), the QKF demonstrats that the obstacle can be detected and tracked more rapidly and accurately.