Alphabet X Unveils Trash-Sorting Robot to Bring Machines Into Daily Life
Alphabet's X lab reveals a robot that learns to sort office garbage, dropping contamination rates from 20% to under 5% as part of a broader push for everyday robotics.
Alphabet subsidiary X reveals its Everyday Robot Project, an ongoing effort to transition robotics from specialized industrial uses to everyday consumer environments. Project lead Hans Peter Brondmo compares the current state of robotics to the computing era of the 1950s and 60s, noting that the primary challenge is teaching machines to navigate unpredictable, mundane elements of human life like moving furniture, changing weather, and shuffled trash bins.
Instead of relying on exact programming, the team collaborates with Google AI to teach the robots using simulation and reinforcement learning. This training approach allows the machines to adapt to real-world variables rather than following rigid, pre-programmed instructions, ultimately making them more genuinely useful to ordinary people.
The project highlights a practical application of this learning method: a robot that successfully sorts through office garbage at X's headquarters. By identifying and separating waste correctly, the robot reduces the office's waste contamination rate from around 20% to less than 5%, proving that teachable machines can effectively handle messy, real-world tasks.