Many years ago before I switched to biology, when I was in the AI group in computer science at uni, we were working on a thing called robot rescue. The idea was to be able to have a coordinated, automated first response - tools that figured out where to fight fires, how to route traffic through damaged cities, where to send aid for maximum effect and how to go about searching dangerous terrain like partially damaged buildings without endangering more lives than necessary. It was a fun project, though what were trying to do was a little ahead of the robotic tools that were availiable at the time. Now though, with things like this, I don't think it's impossible. Maybe not even difficult. It would be quite cool to see a project that I once worked on come to pass, thanks to people like Kumar and his students. Very cool in fact.