Visual Servo Control 中的公式笔记

最新推荐文章于 2023-05-05 10:36:53 发布

lilin2015

最新推荐文章于 2023-05-05 10:36:53 发布

阅读量1.3k

点赞数 2

分类专栏：视觉伺服控制文章标签：机器视觉控制

本文链接：https://blog.csdn.net/lilin2015/article/details/51317549

版权

本文探讨了视觉伺服控制的基本公式和高级方法，包括基本定义、问题、经典图像基视觉伺服和位置基视觉伺服。文章指出，通过最小化误差e(t)来实现目标特征s*，并介绍了交互矩阵L及其逆的计算方法。在经典图像基视觉伺服中，讨论了像素点变化与3D点变化的关系，并提出了获取L^的方法。在位置基视觉伺服中，解释了如何利用L进行控制，以达到期望的相机位置和旋转效果。

摘要由CSDN通过智能技术生成

原文：
Visual servo control. I. Basic approaches. Chaumette(2006)
Visual servo control. II. Advanced approaches. Chaumette(2007)

Basic formulas

$s(m(t),a)$
Compute a vector of $k$ visual feature, with image coordinates of interest points $m(t)$ and additional knowledge about the system $a$ (eg. camera intrinsic matrix).
$e(t)=s(m(t),a)-s^*$
The aim of all vision-based control schemes is to minimize an error $e(t)$ . $s^*$ , the desired values of the features, assumed to be constant.
$\textbf{v}_c=(v_c^T,w_c^T)^T$
$v_c$ , the instantaneous linear velocity of the origin of the camera frame.
$w_c$ , the instantaneous angular velocity of the camera frame.
$\overset{.}{s}=L\textbf{v}_c$
$L{\in}R^{k{\times}6}$ , unknown, named the interaction matrix or the feature Jacobian related to $s$ .
$\overset{.}{e}=L\textbf{v}_c$
Derived from (2)(4).
$\textbf{v}_c=-{\lambda}L^+e$
Come true if we assume that $\overset{.}{e}=-{\lambda}e$ , which means the error has an exponential decrease.
$L^+{\in}R^{6{\times}k}$ , the Moore-Penrose pseudo-inverse.
If $L$ is row full rank, $L^+=(L^+L)^{-1}L^T$ .
If $L$ is column full rank, $((L^T)^+)T$ .
If $L$ is not full rank, take SVD : $L=UDV$ . Then, $L^+=VSU^T$ , $S(i,i)= \begin{cases} 0& \text{D(i,i)=0}\\ {\dfrac{1}{D(i,i)}}& \text{D(i,i)!=0} \end{cases}$

Questions

How should $s$ be chosen?
What is the form of $L$ ?
How should we estimate $\hat{L^+}$ ?
What are the performance characteristics of the resulting closed-loop system?

Classical Image-Based Visual Servo

Basic definition
$m=(u,v)$ , the coordinates of the interest points in pixel units.
$a=(c_u,c_v,f,\alpha)$ , the set of camera intrinsic parameters: $c_u$ and $c_v$ are the coordinates of the principal point, $f$ is the focal length, $\alpha$ is the ratio of the pixel dimensions.
$s=\textbf{x}=(x,y)$ , calculate from the 3D point $(X,Y,Z)$ , $\left\{\begin{aligned}x&=X/Z=(u-c_u)/f{\alpha}\\y&=Y/Z=(v-c_v)/f\end{aligned}\right.$
$L$
Relation between 3D point variation and pixel point variation