From 434371fb891c4472adcdd252f33f815b79673977 Mon Sep 17 00:00:00 2001 From: Abdelhak Bougouffa Date: Wed, 16 Jun 2021 18:05:32 +0200 Subject: [PATCH] Fix a mistake --- chapters/nonlinearOptimization.tex | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) diff --git a/chapters/nonlinearOptimization.tex b/chapters/nonlinearOptimization.tex index e5a203e..ab89073 100644 --- a/chapters/nonlinearOptimization.tex +++ b/chapters/nonlinearOptimization.tex @@ -30,7 +30,7 @@ \subsection{From Batch State Estimation to Least-square} \end{array} \right. . \end{equation} -Through the knowledge in lecture~\ref{cpt:4}, we learned that $ \mathbf {x} _k $ is the pose of the camera, which can be described by $ \mathrm {SE} (3) $. As for the observation equation, we have already explained in lecture~\ref{cpt:5} that it is just the pinhole camera model. To give readers a deeper impression of them, we may wish to discuss their specific parameterized form. First, the pose variable $\mathbf {x} _k $ can be expressed by $\mathbf {T} _k \in \mathrm {SE} (3) $. Second, the motion is related to the specific form of the input, but there is no particularity in visual SLAM (should be the same as ordinary robots and vehicles). We will not talk about it for now. The observation equation is given by the pinhole model. Assuming an observation of the road sign $ \mathbf {y} _j $ at $ \mathbf {x} _k $, corresponding to the pixel position on the image $ \mathbf {z} _ {k, j} $, then, observe The equation can be expressed as: +Through the knowledge in lecture~\ref{cpt:4}, we learned that $ \mathbf {x} _k $ is the pose of the camera, which can be described by $ \mathrm {SE} (3) $. As for the observation equation, we have already explained in lecture~\ref{cpt:5} that it is just the pinhole camera model. To give readers a deeper impression of them, we may wish to discuss their specific parameterized form. First, the pose variable $\mathbf {x} _k $ can be expressed by $\mathbf {T} _k \in \mathrm {SE} (3) $. Second, the motion is related to the specific form of the input, but there is no particularity in visual SLAM (should be the same as ordinary robots and vehicles). We will not talk about it for now. The observation equation is given by the pinhole model. Assuming an observation of the road sign $ \mathbf {y} _j $ at $ \mathbf {x} _k $, corresponding to the pixel position on the image $ \mathbf {z} _ {k, j} $, then, the observation equation can be expressed as: \begin{equation} s \mathbf{z}_{k,j}= \mathbf{K} (\mathbf{R}_k {\mathbf{y}_j}+\mathbf{t}_k), \end{equation} @@ -851,4 +851,4 @@ \section*{Exercises} \item Read Ceres' tutorials (\url{http://ceres-solver.org/tutorial.html}) to better understand its usage. \item Read the documentation that comes with \textit{g2o}, can you understand it? If you still can't fully understand it, please come back after lectures~\ref{cpt:backend1} and~\ref{cpt:backend2}. \item[\optional] Please change the curve model in the curve fitting experiment, and use Ceres and \textit{g2o} to optimize it. For example, write an example with more parameters and more complex models. -\end{enumerate} \ No newline at end of file +\end{enumerate}