Reconstructing 3D environments from common pictures and videos is one of the fundamental steps towards realising augmented reality and automating transportation. This is known as the simultaneous localisation and mapping (SLAM) problem.
The efficiency of the algorithms is of paramount importance in order to implement the reconstruction on large scale environments at sub-cm precision. Algebraic methods and parallel computing can be used to enable new methods towards a new range of possible applications.
Within Scape Technologies, I research on Bundle Adjustment, that is the algorithm comparing a selection of features in pictures and use them to optimise the triangulation of the scene geometry.