search

extr15 / slam_note

Some notes about SLAM, mainly formula derivation
32 stars 3 forks source link

F_delta矩阵求解？ #2

Open LeoGo2017 opened 6 years ago

LeoGo2017 commented 6 years ago

hi，你好，兄弟知道okvis代码中imu误差传播协方差矩阵F_delta矩阵是如何得到的吗？okvis理解中此处省略了。

// covariance propagation
Eigen::Matrix<double, 15, 15> F_delta = Eigen::Matrix<double, 15, 15>::Identity();
// transform
F_delta.block<3, 3>(0, 3) = -okvis::kinematics::crossMx( acc_integral_ * dt + 0.25 * (C + C_1) * acc_S_true * dt * dt);
F_delta.block<3, 3>(0, 6) = Eigen::Matrix3d::Identity() * dt;
F_delta.block<3, 3>(0, 9) = dt * dv_db_g_ + 0.25 * dt * dt * (C * acc_S_x * cross_ + C_1 * acc_S_x * cross_1);
F_delta.block<3, 3>(0, 12) = -C_integral_ * dt + 0.25 * (C + C_1) * dt * dt;
F_delta.block<3, 3>(3, 9) = -dt * C_1;
F_delta.block<3, 3>(6, 3) = -okvis::kinematics::crossMx( 0.5 * (C + C_1) * acc_S_true * dt);
F_delta.block<3, 3>(6, 9) = 0.5 * dt * (C * acc_S_x * cross_ + C_1 * acc_S_x * cross_1);
F_delta.block<3, 3>(6, 12) = -0.5 * (C + C_1) * dt;
extr15 commented 6 years ago

这部分没有理解，印象中改成其他形式，影响不大。