Skip to Content
MIT Technology Review

Google is closing a bipedal-robot project, but the robot revolution is far from over

Google’s parent company Alphabet has decided to close down an ambitious project to develop bipedal robots, created with the acquisition of a Japanese company called Schaft in 2013. 

Robots rising: It’s the latest sign that the robot revolution will take longer than the company originally expected—Alphabet sold off another of its most prominent robot acquisitions, Boston Dynamics, to Softbank last year. But it may also signal a plan to focus on more practical applications, such as manufacturing and picking using conventional robotic arms.

Fancy footwork: Schaft’s robot was extremely advanced. A spin-out from the lab of Hirochika Inoue at the University of Tokyo, it featured innovative actuators, cooling systems, and control software. In fact, the company’s original machine won the first DARPA Robotics Challenge, held in 2013, demonstrating remarkable agility and sure-footedness in several daredevil tasks (like, err, climbing up ladders). The video above shows it in action.

Upside downturn: After the closure of Rethink Robotics, a prominent company working on collaborative robots, it’s tempting to see a downturn in the robot market. But this isn’t the case at all, as worldwide sales of industrial robots show. Schaft’s bot was withdrawn from the second challenge, but the struggles all the competitors experienced throughout simply showed how far these advanced, bipedal robots are from being commercially viable. 

Spending spree: Google acquired several prominent and promising robotics companies back in 2013 as part of a project, known internally as Replicant, led by Andy Rubin, the executive behind the Android mobile operating system. Rubin left the company last year, in controversial circumstances, and his robotics project has floundered since then.

Robo-reality check: Alphabet was right to recognize that robotics would experience a renaissance, but the Replicant effort was too ambitious in the technology it was bringing together. Bipedal robots capable of working in unstructured environments could turn out to be incredibly important, but research in that area is probably better suited to academic labs.

Machine visions: Alphabet is still betting big on robotics. In fact, the company has several important projects in the works, including an effort to have industrial robots learn to perform complex tasks by working in parallel. This is just one of many commercial efforts to combine existing robot hardware with cutting-edge machine learning—something that is very difficult but could, in time, fundamentally change the robotics industry.