融合手机和无人机多传感参数的运动目标跟踪方法

Fusion Sensing Data of Phone and UAV for Motion Tracking

摘要:

针对基于图像的无人机运动跟踪方法存在因图像退化带来的错检和漏检问题,提出一种基于手机和无人机多传感器数据融合的运动目标跟踪方法。将手机IMU(Inertial Measurement Unit,惯性测量单元)数据与无人机的IMU和图像数据作为扩展卡尔曼滤波的输入,其中IMU数据用于滤波器的状态估计,并通过将ORB(Oriented FAST and Rotated BRIEF)方法得到的运动目标图像坐标作为卡尔曼滤波的测量更新部分,再将扩展卡尔曼滤波之后的数据用于校正状态估计,进一步提高无人机运动目标跟踪的准确性。设计实验通过实测数据集来模拟无人机跟踪场景,验证该方法的可行性。实验表明,采用多传感器数据融合的无人机运动目标跟踪方法能够达到0.67m的定位误差,相比于基于图像的方法的精度高,验证了该方法的有效性。

Abstract:

The motion tracking method based on image has wrong detections and observation dropouts due to the image degradation. To solve this problem, the motion tracking method based on fusion sensing data of phone and UAV is proposed. The phone’s and UAV’s IMU data and the camera’s image data are input for EKF(Extended Kalman Filter). Besides, IMU data is used for estimating the system state while the moving target’s coordinate calculated by ORB method, the coordinate is used for EKF measurement update, then correct the state estimate by the data calculated from the EKF. Finally, the feasibility of this method is verified by using the real datasets that simulate the UAV tracking scenario. Experimental results demonstrated this method’s superior accuracy on UAV platform compared to the single camera vision tracking method.