wangwen-believer / dr-slam Goto Github PK

您好，我最近在阅读基于曼哈顿假设的slam框架，作者在论文中都会提及到：使用固定的旋转矩阵R来最小化关于特征的误差函数，进而求解平移t。这部分代码出现在TranslationOptimization（）中。但是对此我有疑惑：在优化的过程中旋转矩阵是变化的，没有固定，故请教是证明回事？感谢！

Hi @WangWen-Believer thanks for your sharing code.

any paper published?
geometric information (parallel or perpendicular) restraint will be used to optimize the point or line in the map? this will get a more accurate map rather than just optimizing the camera pose?

many thanks!