Professor Pieter Abbeel of the University of California at Berkeley today announced the latest development of the robot BLUE in the Berkeley Robot Learning Lab. The robot features low cost (less than $5,000), AI-based control, and can perform everyday activities familiar to humans in an unstructured environment, such as stacking a towel.
Pieter Abbeel is a renowned robotics and machine learning expert in the field. He is currently a professor of electrical engineering and computer science at the University of California at Berkeley, co-director of the Berkeley Artificial Intelligence Laboratory (BAIR), and the Berkeley Robot Learning Lab (UC Berkeley’s Robot Learning Lab). )director. The BLUE robot was developed at the Berkeley Robotics Learning Lab and produced by Berkeley Open Arms. According to Lei Feng Network AI Technology Review, in the past few years, Professor Pieter Abbeel revealed in his media reports that his team is developing low-cost robots based on AI, and now the research and development results have finally come to us.
BLUE robot paper promotion photo
Professor Pieter Abbeel described it as: “BLUE is a low-cost, safe, and capable robot with an AI concept in its original design.” Specifically, more than 15 researchers from different fields in the Berkeley Robot Learning Lab developed a low-cost Quasi-Direct Drive (QDD) controller for BLUE robots to build a complete design paradigm. Unconstrained automation can be achieved; robots can support popular AI-based control methods; they also consider the robot’s own design and robot production process in detail to reduce its cost. BLUE robots can learn in an unstructured environment, have the flexibility to perform many different tasks, and interact with humans.
In terms of technical parameters, the BLUE robot has two arms of similar human size and similar motion, each arm has 7 degrees of freedom and the load is 2kg. The robot’s design pose control bandwidth is 7.5Hz, and the repeat positioning accuracy is no more than 4mm. Such motion characteristics have approached or even exceeded the level of human operators. The robot can learn from the human evolution (Learning from Demonstration) or intensive learning.
The BLUE robot is available for sale, so the Berkeley Robot Learning Lab considered the manufacturing cost of the robot during development and commissioned it for Berkeley Open Arms for subsequent mass production. For the price of end-users after mass production, they are estimated to be less than $5,000. The design, manufacture and pre-testing of the robots have been completed, and now they are continuing the stress testing and mass production preparation of the robots. They hope that other researchers will buy and enjoy the beauty of the BLUE robot.
In order to facilitate the communication of robotic technical details with other researchers, the BLUE robot team wrote a paper “Quasi-Direct Drive for Low-Cost Compliant Robotic Manipulation” in the paper, the design conditions of the BLUE robot, the design and implementation of the robot arm meeting the above conditions, the physical property evaluation of the robot arm, the manufacturing process design and the production cost analysis are described in detail. Very comprehensive. The paper was also submitted to the IEEE International Conference on Robotics and Automation (ICRA) and was accepted.
The robot homepage can be found at http://rll.berkeley.edu/blue/ and can also be purchased in advance.