CodeSLAM — Learning a Compact, Optimisable Representation for Dense Visual SLAM

Abstract

While each keyframe with a code can
produce a depth map, the code can be optimised efficiently jointly with pose variables and together with the codes of overlapping keyframes to attain global consistency
对关键帧进行编码，跟姿态和重叠的关键帧一起优化

contributions

• The derivation of a compact and optimisable representation of dense geometry by conditioning a depth autoencoder on intensity images.
• The implementation of the first real-time targeted
  monocular system that achieves such a tight joint optimisation of motion and dense geometry

个人理解

• 使用光度信息进行code的自动编码，用RGB-D数据进行训练，然后得到每个code的深度，属于直接法的一种
• SLAM系统：利用滑动窗口的方式，重建利用了SFM