The AI-enabled Blue robot can safely master a task while working alongside humans.
Blue was developed by a team from the University of California Berkeley, and can learn new tasks via reinforcement learning (trial and error). Since making an error can damage a robot—and any human working near it—Blue features flexible plastic arms with rounded edges that can be easily pushed aside if they stray too far out of bounds.
According to graduate student David Gealy, "Essentially, we can get more out of a weaker robot. And a weaker robot is just safer. The strongest robot is most dangerous. We wanted to design the weakest robot that could still do really useful stuff.”