It already is. There is github code example of what the paper tackles and the community is putting together a project where a tiny agent is learning the structure of a map
Look up the 2d object learning thread. Exciting stuff
Gridcells paper code: https://github.com/numenta/htmpapers/tree/master/frontiers/location_in_the_neocortex_a_theory_of_sensorimotor_object_recognition_using_cortical_grid_cells