Introduction

SLAM main component:

Sensor data retrievation

Visual Odoemtry

Front End: feature abstract and matching

Back End: optimiation

Loop detection / loop closure

Goal: Location and Mapping

Types:

Monocular SLAM / stereo / Laser / distance odometry /

dense/sparse IMU

Scale Ambiguity

accumulating drift

movement equation vs observeration equation

Programming

cmake minimum version / library / target

c++

cmake tutorial

1. TheErk

Movement in 3d space

1. skew-symmetric matrix A(T) = -A :
   a^: right upper corner: a3 a2 a1

2. direction cosine matrix –> rotation matrix –> transform matrix

   a1 = R12 * a2 + t12

3. rotation matrix / angle-axis or axis-angle / rotation vector - theta

   trace(matrix) = 1 - 2 cos(theta)

   rotate ‘rotation vector’ does not change the ‘rotation vector’, thus R * n = n

   euler: yaw-pitch-roll / gimbal lock

   quaternion

   ij = k, guess i - 90 and j -90 whether equal k - 90: no. acutually i - 180

4. useful link:

understand euler angle and rotation matrix (also angle-axis)

understand quaternion and its interpolate method 旋转的表示

回顾三角函数计算

Gimbal Lock

Rotation and Gimbal Lock