Not for general public driving, of course, but for fleets of trucks and excavators that operate globally 24/7.
At various sites they are already operating driverless equipment using tech developed in house.
For what it's worth, oil & gas & mineral exploration work drove a large part of the civilian development of tools like Google Earth / Google Maps in the decades prior to the first public offerings by Google (who acquired their tech from others (the chaps from Sydney for Maps, Keyhole for Earth)).
I still see it as orders of magnitudes off in complexity. You're talking about a system for autonomously controlling every motor vehicle in the U.S., or even the world -- that's effectively the end game. Google is among the handful of companies that has the computing infrastructure to have a "pulse" on the world at all times, and within those companies they are heads and shoulders above the rest in terms of being able to implement large scale, distributed information processing algorithms with low latency at that scale. It's a very rare, maybe completely unique cross section of capabilities. Google is better suited to solving this problem in many ways than any other problem they're currently solving.
The point is the capabilities I'm talking about have nothing to do with cars and everything to do with computing. Pointing to another company doing autonomous vehicles as a counterexample is kinda-sorta missing my point.
> You're talking about a system for autonomously controlling every motor vehicle in the U.S.,
Not a (single) system, rather one such system per car - standard protocols for inter system comms are yet to arrive and each vehicle needs to react to local changes such as pedestrians w/out reference to an external global computing infrastructure.
It's not a big data, massively interconnected problem at all - the elegance comes from local smart behaviour - even for broader problems like optimal pooling.
Others can enter the driverless vehicle market and there are companies like Akamai Technologies that could provide any required interconnect. Google is a player, but by no means the only player.
It's both. Its hyperlocal real-time understanding and eye-in-the-sky big data analysis to keep everything running smoothly. Think about where you end up 10-15 years after this technology comes out: "cars" driving on multi-level, many-laned highways travelling 150-200mph a few meters apart, traffic flow coordinated to optimize energy usage and minimize time to destination. It's a hard problem and not solvable via local agents alone.
> It's a hard problem and not solvable via local agents alone.
Oh, Really?
Ants seem to do okay.
Following your, umm,, solution of critical external hyper networked global oversight it seems there'd be the mother of all traffic accidents if the servers went out.
Biologically emergent local behaviour seems wonderfully adept at converging on globally optimal solutions via local agent actions. Perhaps you're right, maybe Google is the only company capable of replicating the efficiencies of termite mounds.
Not for general public driving, of course, but for fleets of trucks and excavators that operate globally 24/7.
At various sites they are already operating driverless equipment using tech developed in house.
For what it's worth, oil & gas & mineral exploration work drove a large part of the civilian development of tools like Google Earth / Google Maps in the decades prior to the first public offerings by Google (who acquired their tech from others (the chaps from Sydney for Maps, Keyhole for Earth)).