Teach a robot to open a door, and it should open up possibilities for a lifetime. Not so for one of Alphabet’s newest subsidiaries, Everyday Robots. A little over a year after graduating from Alphabet’s X Moonshot Lab, the team, which has trained over a hundred one-armed robots on wheels to wipe down cafeteria tables, separate trash and recycling, and, yes, open doors, is part of the expanding A spokeswoman confirmed that budget cuts were made to the Google mother company.
“Everyday Robots will no longer be a separate project within Alphabet,” said Denise Gamboa, director of marketing and communications at Everyday Robots. “Some of the technology and some of the team will be consolidated into existing robotics efforts within Google Research.”
The robotics venture is the latest failed bet for X, which also spun internet-radiating balloons (loon) and electricity-generating kites (makani) for the past decade before deciding them too commercially unviable to stay afloat. Other one-off X projects like Waymo (development of autonomous vehicles) and Wing (testing of drones for food delivery) continue as companies within Alphabet, though their financial prospects continue to be marred by regulatory and technological challenges. Like Everyday Robots, these ventures used novel technologies that showed impressive promise in tests but weren’t rock-solid reliable.
Everyday Robots emerged from the rubble of at least eight robotics acquisitions by Google a decade ago. Google co-founders Larry Page and Sergey Brin expected machine learning to reshape robotics, and Page specifically wanted to develop a consumer-oriented robot, says a former employee involved anonymously at the time, to discuss internal considerations. By 2016, they hired software entrepreneur Hans Peter Brøndmo to lead a project then known as Help (and later temporarily Moxie) to use machine learning to create robots that could do routine tasks and adapt to different environments, according to the source.
The team set up arm farms and playpens where a fleet of robots repeated the same task — like sorting garbage — for months. It was a brute force attempt to generate data to train a machine learning model that could then embody the robots with the know-how needed to use their cameras, arms, wheels, and finger-like grips to to interact with the world around them. The novelty saves engineers from the traditional approach in robotics of programming specific instructions for the machines to follow for each small potential scenario. The idea largely worked for initial tasks. Google has had its fleet of Everyday Robots help clean the search giant’s dining rooms and scavenge for messy conference rooms during the pandemic.
Over the past year, Everyday Robots demonstrated further advances in Google AI researchers. The project integrated a large language model similar to the underlying ChatGPT into the robotic system, allowing the mechanical helper to respond to hunger, for example, and fetch him a bag of chips. But Google and Everyday Robots emphasized at the time that a roving on-call butler was far from available to consumers. Variations that appear trivial to humans, such as the type of lighting in a room or the shape of the chip bag, could lead to malfunctions.
From the start, Everyday Robots struggled with whether its mission was to conduct advanced research or bring a product to market, says the former employee. It employed up to 200 people, including people who oversaw customer operations, taught robots to dance, and worked to perfect the design. Each of its robots is likely to cost tens of thousands of dollars, robotics experts estimate.
That spending was too much for Alphabet, whose more speculative “other bets” like Everyday Robots and Waymo lost about $6.1 billion last year. Alphabet’s overall profit fell 21 percent to $60 billion last year as spending on Google ads fell and activist investors called for the company to make cuts. On Jan. 20, Alphabet announced it would lay off about 12,000 workers, 6 percent of its workforce. Everyday Robots was one of the few projects that got dissolved.