In 2017, X quietly teased its robotics efforts as being heavily machine learning-focused. Alphabet’s moonshot factory is now finally detailing the project and its mission to build Everyday Robots.
The Everyday Robot Project wants to teach “robots how to help us in our everyday lives.” It comes as the team at X believes that “we are much closer than ever to robots becoming mainstream” thanks to machine learning advancements, sophisticated sensor technology, and low-cost hardware.
As we’ve previously detailed, this program has a long history:
Despite the huge waves Google made in 2013 by acquiring several robotics companies and 300 engineers, there is not much to show for it. A majority of these companies — but not Boston Dynamics — coalesced into a group known as Replicant before being absorbed into X. In late 2015, X head Astro Teller told employees they would be reassigned if robotics didn’t end up helping Google’s efforts.
In that time, X has come to the conclusion that “you have to teach machines to perform helpful tasks; you cannot program them.”
For robots to be useful in everyday environments we need to move away from painstakingly coding them to do specific and structured tasks in exactly the right way at exactly the right time.
According to X, robots are not only able to learn from repeating simple tasks, but then be taught new capabilities. The company is currently testing this with robots that sort through trash and make sure they are placed in the proper recycling bin at the X office in Mountain View. Sorting through thousands of pieces of trash, the machines helped reduce waste contamination levels from 20% to less than 5% in recent months. There are two takeaways:
- First, they show that we’re able to create a robotic system that integrates all of the robot’s capabilities to do something genuinely useful: reducing the contamination levels to less than 5% successfully diverts waste from going to landfill.
- Second, they prove that it’s possible for robots to learn how to perform new tasks in the real world just through practice, rather than having engineers “hand code” every new task, exception, or improvement.
Ultimately, the X Everyday Robots Project predicts that it “will be will be years before the helpful robots we imagine are here.”
FTC: We use income earning auto affiliate links. More.
Comments