Yup, I very much liked the broader approach of solving the general problem, but this actually seems to be doing better.

The general problem has insane amounts of edge cases, and is basically trying to do a pole-vault-height high jump; if you succeed, you own the world (or at least can automatically drive anywhere on it), but until then, you and your customers have lots of problems.

Meanwhile, the approach of map-everything-down-to-the-centimeter and keep it updating is making real practical progress. The question is whether it'll ever get out of cities? While storage is constantly getting cheaper, but is it feasible to map and update the maps sufficiently for a cross-country trip?

Still, there's an awful lot of city territory that can be covered with genuinely useful service.