Facebook has lifted the curtain on its robotics research. The company detailing three of its robotics projects. Each focuses on finding ways for robots to teach themselves from experience rather than data prepared for training purposes.
“This work will lead to more capable robots,” Facebook wrote, “but more important, it will lead to AI that can learn more efficiently and better generalize to new applications.”
One project centers on enabling a six-legged robot to teach itself to walk. Another explores how equipping robots with “curiosity” could improve the learning process, while a third focuses on enabling robots to learn through tactile sensing, or “touch.”
“The real world is messy, it’s difficult,” Roberto Calandra, a research scientist in Facebook’s AI division, told. “The world is not a perfect place; it’s not neat.
So the fact that we are trying to develop algorithms that work on real robots will help to create [AI] algorithms that, generally speaking, are going to be more reliable, more robust, and that are going to learn faster.”