AirDos summary #

1. Introduction #

Elimination strategy can handle environments with a small number of dynamics, but cannot address challenging cases, where dynamic objects cover a large field of view as in Fig. 1(a).

Joint Optimization Method #

1. a rigidity constraint, which assumes that the distance between any two points located on the same rigid part remains constant over time.假设两点之间的距离在连续图像之间不变。

$$e_r=||_lp_k^i-{}_lp_k^j|-s^{ij}|.$$

1. a motion constraint, which assumes that feature points on the same rigid parts follow the same 3D motion.假设同一刚体上的特征点具有相同的运动，这里只有位移，没有旋转，所以对加法是封闭的，不用做什么处理。

$$e_m=||l\bar{p}{k+1}^i-lT{l}\bar{p}_k^i||.$$

This allows us to build a 4D spatio-temporal map including both dynamic and static structures.(这句话没看懂，什么叫时空地图，是观测结果的表示吗？)

G2O基础 #

1. 顶点 (Vertex): 顶点代表优化问题中的状态变量。在g2o中，需要定义每种类型的顶点类，继承自g2o::BaseVertex。实现setToOriginImpl()和oplusImpl()方法。

2. 边（Edge） 边代表状态变量之间的约束或观测。在g2o中，定义边类，继承自g2o::BaseBinaryEdge或g2o::BaseUnaryEdge。实现computeError()方法。

   

3. 添加边和点

   

4. 优化