Robots Learn Simple Human Tasks in Watertown, Massachusetts – Alyssa Andrews Reports for WBZ-TV
- Robots are learning to do simple human tasks at a factory in Watertown, Massachusetts, where the largest AI robot data factory in the United States is being operated...
- The facility, described by Gruenstein as a "kindergarten for robots," focuses on teaching machines basic physical tasks such as picking up individual items, placing them into boxes, and...
- Unlike industrial robots in car factories that repeat the same motion daily, these robots are being trained to handle variability in real-world environments, a challenge Gruenstein notes stems...
Robots are learning to do simple human tasks at a factory in Watertown, Massachusetts, where the largest AI robot data factory in the United States is being operated by Tutor Intelligence, a company co-founded by MIT graduates Josh Gruenstein and Alon Kosowsky-Sachs.
The facility, described by Gruenstein as a “kindergarten for robots,” focuses on teaching machines basic physical tasks such as picking up individual items, placing them into boxes, and attempting to fold laundry—activities that require adaptability and fine motor skills not yet mastered by current robotic systems.
Unlike industrial robots in car factories that repeat the same motion daily, these robots are being trained to handle variability in real-world environments, a challenge Gruenstein notes stems from the lack of large-scale data for physical human tasks comparable to the text and image datasets used to train AI models like ChatGPT.
According to Gruenstein, the core objective is to develop robots capable of learning any behavior in any situation by creating the necessary data through direct observation and repetition, since no existing dataset contains sufficient examples of diverse physical human actions for AI to learn from.