Within UC Berkeley’s Robot Learning Lab, we have been developing AI for robotic learning and control, seeking to answer the question:

“How can AI improve the capabilities of existing robots?”

For Project BLUE, we instead turned our attention towards developing robots for AI-based control, seeking to answer the question:

“How can AI change the robot design paradigm?”

After a three year effort across a multidisciplinary team of more than 15 researchers, we’ve designed, built, and tested BLUE - the Berkeley robot for Learning in Unstructured Environments. BLUE is a low-cost, high-performance robot that is intrinsically safe, developed from the ground up with ever-increasing Artificial Intelligence capabilities in mind.

In a paper to be presented at the IEEE International Conference on Robotics and Automation in Montreal in May, we describe the key design criteria, trade-offs, and actuation performance for BLUE.

Next up for the project is continued stress testing and ramping manufacturing. From there, the goal is to get these affordable robots into as many researchers’ hands as possible.

Prototypes of BLUE have already been shipped to early adopters through the organization Berkeley Open Arms. If you’re interested in getting a BLUE robot for yourself, fill out our interest form to get updates as well as priority access!