In this article, we described about the UWB-IMU integrated system for indoor pose estimation of unmanned aerial vehicles (UAVs). Based on the extended Kalman filter (EKF) algorithm, the pose of the UAV and position of UWB anchors can be estimated, but it can only approximate a uni-modal probability distribution function. To estimate the UAV’s two dynamic models (linear motion and rotation) properly, we propose a multi-modal algorithm which is calculating weighted sum of Gaussians of two different states. We simulated this algorithm with indoor motion tracking system.