I used vins-fusion to run the data set package recorded by my drone (including Stereo data + angular velocity, gradient, and quaternary data). The imu data was fused by a script. When only Stereo data was run to estimate positioning, the observations were roughly Correct; but when running Stereo + imu with vinsfusion, the warehouse position goes straight to infinity in the negative direction of the z-axis at the beginning, and the position cannot be estimated. that's why？can someone help me ,thinks.
I used vins-fusion to run the data set package recorded by my drone (including Stereo data + angular velocity, gradient, and quaternary data). The imu data was fused by a script. When only Stereo data was run to estimate positioning, the observations were roughly Correct; but when running Stereo + imu with vinsfusion, the warehouse position goes straight to infinity in the negative direction of the z-axis at the beginning, and the position cannot be estimated. that's why？can someone help me ,thinks.
我用vinsfusion运行自己的无人机录制的数据集包（包含双目数据+角度速度、梯度、四元数据），imu数据由一个脚本融合出来，当仅运行双目数据估计定位时，观测大抵正确；但是用vinsfusion运行当运行双目+imu的时候，刚开始仓库位置就直冲z轴负向无限大，估计不出位置。这就是为什么呢