vslam

This is a coding practice of bundle adjustment for visual slam, i.e., an Levenberg-Marquardt optimizer with only the usage of Eigen and Sophus. When solving the equation, the operations are manipulated by block-matrix. Some code snippets are learned from the following reference:

1. https://blog.csdn.net/bdl295/article/details/90373529
2. https://github.com/ydsf16/vslam
3. https://github.com/gaoxiang12/slambook