Skip to content

Commit

Permalink
Merge pull request #49 from hnqiu/master
Browse files Browse the repository at this point in the history
Fix: duplicate text & a typo in lieGroup.tex
  • Loading branch information
gaoxiang12 committed Sep 8, 2021
2 parents 4c8180c + 22dc81c commit c5b5528
Showing 1 changed file with 1 addition and 2 deletions.
3 changes: 1 addition & 2 deletions chapters/lieGroup.tex
Original file line number Diff line number Diff line change
Expand Up @@ -387,8 +387,7 @@ \subsection{Derivative on $\mathrm{SO}(3)$}
\mathop {\min }\limits_{\mathbf{T}} J(\mathbf{T} ) = \sum_{i=1}^{N} \left\| {\mathbf{z}_i - \mathbf{Tp}_i} \right\|^2_2.
\end{equation}

Because of the noise, the real observed data is not absolutely the same as the one we computed from the observation model, so we can calculate the error of predicted observation with the real one:
To solve such an optimized problem (which is a least square problem), we need to calculate the derivative of $J$ by $\mathbf{T}$. We leave the least square problem to the next section. Here we just want to clarify that we normally have some functions that have rotations or transforms as their variables. We have to adjust those rotations or transforms to find a better/best estimation. But, as we mentioned before, since $\mathrm{SO}(3)$ and $\mathrm{SE}(3)$ do not have a well-defined addition (they are just groups), so the derivatives cannot be defined in their common form. If we treat the $\mathbf{R}$ or $\mathbf{T}$ as common matrices, we have to introduce the constraints into our optimization.
To solve such an optimization problem (which is a least square problem), we need to calculate the derivative of $J$ by $\mathbf{T}$. We leave the least square problem to the next section. Here we just want to clarify that we normally have some functions that have rotations or transforms as their variables. We have to adjust those rotations or transforms to find a better/best estimation. But, as we mentioned before, since $\mathrm{SO}(3)$ and $\mathrm{SE}(3)$ do not have a well-defined addition (they are just groups), so the derivatives cannot be defined in their common form. If we treat the $\mathbf{R}$ or $\mathbf{T}$ as common matrices, we have to introduce the constraints into our optimization.

However, from the perspective of Lie algebra, since it consists of vectors, it has a good addition operation. Therefore, there are two ways to solve the problem of derivation using Lie algebra:

Expand Down

0 comments on commit c5b5528

Please sign in to comment.