Alphabet, the parent company of Google, has announced that Project Daily Robots – a team within Experimental X Labs dedicated to creating a general-purpose educational robot – has moved some of its model machines from the lab to the research giant’s San Francisco Bay Area headquarters to perform some light-duty tasks.
We are now operating a fleet of more than 100 prototype robots that independently perform a range of useful tasks around our desks, chief robotics officer of the Everyday Robotics Project, Hans-Peter Brondmow, said in a blog post. The same robot that sorts trash can now be fitted with a squeegee to sweep tables and use the same clutch that can teach it to hold cups to open doors.
These robots in question are essentially arms on wheels with a multi-purpose clutch at the end of a flexible arm attached to a central tower. There is a head on top of the tower with cameras and sensors for machine vision and what looks like a rotating lidar unit on the side, presumably for navigation.
As Brondmo points out, these bots were first seen sorting recycling when Alphabet launched the Everyday Bots project team in 2019. The big promise of the company (as well as many other startups and competitors) is that machine learning enables bots to operate in environments Not organized like homes and offices.
Companies can now build machines that can perform repetitive tasks in the factory. But it’s hard to get machines to repeat simple tasks like cleaning the kitchen or folding clothes
You may have seen robots from Boston Dynamics perform back somersaults. But you haven’t seen them take out the trash, because it’s hard to get machines to handle things they’ve never seen before in a new environment (something humans do every day). And this is the problem that Alphabet wants to solve.
The new rendering is an update on how the company is progressing toward building autonomous machines and testing them in real-world offices.
Alphabet aims to build robots that can learn on their own. Instead of having to meticulously program it to solve a single problem in just one environment.
Most robots still operate in specially designed and structured environments, and the tasks they complete are very specific. The robots are carefully programmed to perform those tasks in exactly the right way, at the exact right time. However, this approach will not work in the messy complex spaces of our daily lives.
Robots can transfer learning from the virtual world to the real world. Complex tasks (eg, opening a door) can now be learned in a day. And that’s a higher success rate than it was five years ago. When assimilating things took the equivalent of four months.
And using a combination of machine learning techniques such as reinforcement learning, collaborative learning and demo learning. Robots are steadily gaining a better understanding of the world around them and are becoming more adept at performing everyday tasks.
The Everyday Robots project team envisions a future in which machines will work alongside humans to simplify everyday tasks. Google’s long-term goal is to enable older adults to maintain their independence for longer.