Hello,
I'm working on mapping the car's environment by updating an occupancy
grid. An occupancy grid dicretizes the 3D space in small grid elements
(voxels). A grid element contains informations about the space it's
representing.
I need to map a space of the dimensions 50m * 5m * 3m with grid
elements of size 1cm * 1cm * cm (=3D 750 000 000 grid elements).
My question is how to implement such a data structure in an efficient
way. I need to access fast by indizes. Access by iterators is not
needed. The implentation has to be dynamic because of the iterative
mapping process.
My idea was to store all voxels in a long std::vector<double> and let
pointing a kd-tree's leafs on the vector elements. But this would have
the drawback that the kd-tree has to be reorganized during the update
process to avoid a degeneration to a linear list.
Does anybody has an other idea? Or exemplary code?
Best regards,
Marco K=F6rner
