The field of robotics is zooming past the lumbering assembly line machines of today, quite literally with machines that can run, jump and climb. But the new breed of robots uses vast amounts of data about the world in addition to their high-tech motors, springs and sensors.
That’s likely why Google (GOOG) is getting into the business. The search giant last week bought Boston Dynamics, marking its eighth acquisition of a robotics company in the past six months, the New York Times reported over the weekend. But while Google’s previous acquisitions were companies that made bits and parts of robots, the Boston Dynamics deal makes clear that Google’s true ambition is human-like robots interacting with ordinary people.
“The only reason to buy this company is to make complete androids, systems that can walk around on our sidewalks and right up to our homes," says Illah Nourbakhsh, a professor at Carnegie Mellon University and author of the book Robot Futures. Such bots could be incredibly helpful, but also incredibly invasive as they send data and pictures back to Google, he says.
Started by former MIT professor Marc Raibert, Boston Dynamics so far has designed animal-like machines that can carry heavy packs, climb mountains and even run at high speed. The company’s Wildcat robot can gallop at 16 miles per hour, as shown in a video posted to Youtube and viewed over 15 million times.
The Cheetah can run even faster, at an astounding 28.3 miles per hour.
Founder Marc Raibert has said that, since he joined the private sector, he no longer measures his influence by how many other academics cite his work, but intead by how many views his company's videos attract on YouTube. He should be able to grab even more attention now that he'll be working for YouTube's owner.
Most of Boston Dynamics' agile beasts were built as military research projects, but the company also worked on more consumer-friendly efforts, such as Sony’s Aibo robot dog. A future product using Boston Dynamics technology could be a set of assistive legs for wheelchair-bound people, Nourbakhsh says. Also likely, as many have speculated, are delivery robots and other machines that will have human shape to best fit into the human world, he says.
With constant connections to computer servers over the Internet, the robots will be able to act as two-way connections between the digital and physical worlds. Such robots will get cues on how to behave with knowledge of a person’s preferences as collected by Google and, at the same time, send data back to Google of observations in the real world.
Reams of data are also the key to Google’s self-driving cars. Building robotic cars that navigated only by analyzing their surroundings in real time would be extremely difficult. But giving the robotic car detailed maps and annotated pictures simplifies the task immensely, as Google’s head of research Peter Norvig explained in the New York Times Magazine on December 15.
The flip side of data-driven robots is data-collecting robots, and that has some privacy advocates worried about Google’s activities. A robot in the home could have unprecedented access to a family’s activities and preferences.
“Eventually, you might be able to market things with such detail to each individual that it would be nearly impossible to say no,” warns Nourbakhsh. “Then, we have basically become the robots.”