The future according to Google

Self-driving cars, natural language programming and interior mapping: we explore Google's research labs

Barry Collins
11 Jan 2013

The majority of Google’s income is still derived from search, but nobody could accuse the company of not trying to broaden its reach. In 2011, the company spent more than $5 billion on research and development – 12% of its overall revenue. That figure looks set to increase significantly in 2012, too.

Driving forward

The future according to Google

Google’s best-known research project is so far off the beaten track that the company has faced accusations that it’s frittering away billions on a vanity project for its ever-ambitious founders.

Yet Google’s self-driving car is edging closer to the showroom. It’s been approved for use on the roads of Nevada, Florida and California, provided there’s a human in the driving seat to spot hazards missed by the barrage of sensors (touching the steering wheel, pedals or the big red kill switch in the centre console reverts the car to human control). The company is lobbying hard for access to the roads in several other states, too. In fact, the challenge of getting the car on the road is as political as it is technological. Google spent $9 million in the first half of 2012 lobbying US officials in the Department of Transportation and others, according to The Wall Street Journal.

That isn’t to say there aren’t technological problems to solve. Despite a 64-beam laser scanner on the top of the vehicle, a radar in the front grille, and a camera behind the windscreen, the sensors are unable to detect road markings when the road is wet or snowy. Neither is the car able to park itself – which is odd, considering some of today’s more advanced road cars are already capable of automatic parallel parking. Yet if Google can smooth over the bumps, it may find its competitors in the next decade aren’t Facebook or Apple, but Ford and Toyota.

Natural language programming

Other Google research projects are a little closer to familiar territory. Engineers are, for example, working on a new programming language called Sense Everything, Control Everything (SECE), which aims to make it possible for people to automate tasks using natural language commands. The concept will be familiar to anyone who has used the wonderful web service IFTTT (If This Then That). IFTTT allows you to create “recipes” for selected web services, making it possible to automatically change your Facebook profile every time you change your Twitter avatar, for example, or send your Instagram photos to your inbox.

SECE takes it a stage further, allowing users to create “rules” for devices ranging from smartphones to home entertainment systems.
Users could write a script that switches on the hallway light at sunset, or sends them a text message if the GPS sensor in their child’s smartphone detects that they’ve strayed more than a mile from school.

The researchers chose to develop SECE in a natural-English-like language “because it’s more powerful and easy to use than XML and form-based solutions”. The script to turn on the home lights when the sun goes down is no more complicated than:

every sunset
{ homelights on; }

The challenge, of course, will be gaining support from manufacturers and developers, although the researchers have ensured their language has the best chance. SECE is being developed in Java “due to its extensive libraries and support for operating systems”, and agents for services such as Facebook are already in development.

Mapping buildings

Another project gives a glimpse of how the company could generate revenue from its huge investment in Android. The project comprises an app that can create a 2D floor plan or 3D model of the inside of a building, using equipment no more sophisticated than the smartphone’s built-in camera.

Using a concept similar to that used in Google’s Street View cars, the app captures data by having the user stand in the centre of the room and make a 360-degree sweep with a smartphone camera. They also have the option to take photos of particular objects and mark specific features.

The future according to Google

“Realtors and homeowners can use our smartphone application to generate tours and floor plans of houses without the need for any extra equipment,” the research paper claims. Virtual tours of public buildings is another potential application.

However, the team concedes there are technical obstacles to overcome first. GPS doesn’t work well indoors, making it difficult to calculate the user’s exact position as they move from room to room. Even today’s top-end smartphones may struggle with the computational demands of stitching together dozens of images from different rooms.

Read more about: