Housed in a former shopping mall a half mile from Google’s main campus is a facility bursting at the seams with off-the-wall ideas. Employees skateboard and bike through the massive halls. There are dogs.

This is X (formerly Google X), the company’s “moonshot factory” dedicated to solving really, really tough world problems with really, really crazy technology. Headed by Sergey Brin, the lab is working on four major projects today.

Self-Driving Car

By 2020, Google’s self-driving cars will hit the market. That gives X engineers just four more years to perfect their prototype.

Prototype self-driving cars have already traveled more than a million miles without human help, the equivalent of 75 years of driving practice. The cars have tackled Lombard Street’s hairpin turns, the Golden Gate Bridge, and roads around Lake Tahoe.

There are still a few challenges to solve before they can be unleashed on the public, though. The cars haven’t been tested in snow and can’t tell the difference between a rock or a piece of paper, so they swerve to avoid both. They haven’t quite figured out how to avoid pedestrians who are about to step into an intersection: one patent, granted to Google last fall, details ideas like an LCD screen that displays “coming through,” a loudspeaker for verbal warnings, and even a robotic hand that would wave.

Each of the tiny, podlike cars has about $150,000 in equipment inside, so you’re not likely to see one as a personal vehicle right away—think taxis and shuttles first.

Project Loon

Its name implies a touch of madness, but this idea may graduate from X as soon as later this year.

The effort to supply high-speed Internet to underserved areas by launching Internet-connected balloons into the stratosphere is being tested in Sri Lanka this year, following successful previous tests in New Zealand, California, and Brazil. India and Indonesia are reportedly interested.

In addition to the balloons, each as large as a tennis court, with a lifespan of four months, Project Loon includes a device, nicknamed Chicken Little, that can fill and launch the balloons in 30 minutes each. The balloons communicate with each other in the air and shift altitude to catch winds that move them where they need to go.
Makani

Some places are more suitable for wind turbines than others. You need lots of land and wind that consistently blows—and neighbors who won’t complain about the view and kill a project before it starts.

Enter Makani, which generates energy from wind using kites. Yep, you read that right. A specially designed kite in flight generates up to 600 kilowatts, sending the power back down through a tether to earth. That’s enough to power over 100 homes.

Google purchased Makani Power in 2013 and rolled it into X that year. In late 2015, Google posted a number of open positions for Makani—including customer-facing positions like sales engineers—hinting that the prototype is soon to become a real product.

Project Wing

Move over, Amazon. Google’s getting into the drone product delivery business.

Announced in 2014, Project Wing involves autonomous drones, developed by MIT roboticist Nick Roy, that Google says will be able to deliver people’s orders in just a minute or two. This will, according to X director Astro Teller, transform people’s relationship to stuff. Why buy a drill that you use once a year or stockpile batteries if you can rent the drill or buy batteries one at a time?

The technology behind the project has evolved significantly since it was first announced. The team originally developed a delivery system where the drone would lower packages to the ground via winch, but a recently granted patent describes “mobile delivery receptacles”—wheeled boxes on the ground that communicate with and guide the drones—that would accept a package from a drone and move it to a secure holding location.

There are still regulatory hurdles to overcome, but Google still says Project Wing could be making commercial package delivery by drone as early as 2017.

Those are the current four projects in development at X. Others have already “graduated”—like Google Glass—and others have been quietly killed after being deemed too out there. X tried to build a jetpack that proved too energy inefficient, and a tiny hoverboard that didn’t scale up well to larger sizes. X also took a look at building a space elevator—rejected because of a lack of viable construction materials—and teleportation—rejected after concluding the idea violates the laws of physics. But, pointed out journalist Eric Mack in a 2014 article, even those discussions ended up leading to insights into new encryption technologies.

That’s why X encourages ideas and solutions “that sound impossible today, almost like science fiction.” Because you never know where asking the right question might take you.