Meet Blue, the low-cost, human-friendly robot designed for AI

Meet Blue, the low-cost, human-friendly robot designed for AI

Robots can have a habit for super-human strength and precision, but they still struggle with some basic human work - like washing clothes or making a cup of coffee.

Entered Blue, a new low-cost, human-friendly robot imagined and produced by a team of researchers from the University of California, Berkeley. 

Blue was designed to use the recent advances in artificial intelligence (AI) and to use deep reinforcement to learn deep human work, while all affordable and safe enough that every artificial intelligence researcher - and finally everyone can be at home.

Blue Pepper is the brainchild of Abel, who has gone to UC Berkeley Professor of Electrical Engineering and Computer Science, Postdoctoral Research Fellow Stephen McKinley and graduate student David. 

The team hopes that Blue will speed up the development of robotics for the home.

"AI has done a lot for existing robots, but we wanted to design a robot that is right for AI," Abel said. "Existing robots are very expensive, are not safe around humans and so are not even safe around themselves - if they learn through trial and error, they will easily break themselves.

 We want to create a new robot It was right for the AI era. High-precision, sub-millimeter, factory automation age. "

In the last 10 years, Abel has pioneered intensive reinforcement learning algorithms, which helps the robot to be guided by humans like testing and error or puppetry. 

They developed these algorithms using robots manufactured by outsourcing companies that market them thousands of dollars.

Blue's durable, plastic parts and high-performance motors are less than $ 5,000 overall. Its sides, each about the size of the average bodybuilder, are sensitive to external forces - like one hand pushing it away - and rounded edges and the minimum pinch points which avoid catching the stray fingers. Blue arms can be very harsh, such as human flexing, or very flexible, such as human comfort, or anything in between.

At present, the team is building 10 weapons at home to distribute for early adoption. 

Read more: Oppo Reno Series Set to Launch in India on May 28

They are continuing to investigate the stability of Blue and to deal with the formidable challenge of building robots on a large scale, which will be through the UC Berkeley Spinoff Berkeley Open Arms.

 Sign-ups to express interest in the priority access that started on that site today,

"With a low-cost robot, every researcher can have his own robot, and this vision is one of the main driving forces behind this project - more research is being done in the world than being more robots," McKinley said.

From floating statue to the cat to the litter

Robotics has traditionally focused on industrial applications, where robots require strength and accuracy to perform fully repetitive tasks.

 These robots flourish in a highly structured, approximate environment - traditional Americans cry far away from home, where you can get access to children, pets and dirty clothes on the floor.

"We have often described these industrial robots as moving sculptures," said the pastor. 

They are very strict, meaning A to point B and from back to point A completely. 

Read More: World-record quantum computing

But if you ask them to go for a centimeter from the table or wall, they will break into the wall and lock. " Break yourself or break the wall. nothing good. "

If any AI is making mistakes and working in unstructured environments, then these rigid robots will not work only. 

To secure the experiment, Blue was designed to force-control - highly sensitive to external forces, at any time to modify the amount of force always at any time.

"It's really good about the design of this robot that one of the things is that we can make it force-sensitive, good and responsive, or we can choose to be very strong and very tough," he said. "Researchers can adjust how rigorous the robot is, and what kind of hardness - do you want it to feel like jaggery? Do you want to feel like it's spring? A combination of those people? We want the robots to go home and perform in these fast unstructured environments, they are going to need that capability.

To achieve these abilities at a lower cost, the team thought that what Blue needs to do to meet human-centric functions, and what can be done without it. 

For example, researchers gave Blue a wide range of motion - there are pairs in it which can move forward in the same direction of the human shoulder, elbow, and wrist - through which humans can easily teach how to use virtual reality Complete hard maneuvers. But the lack of strength and accuracy of a specific robot in agile robot weapons.

"We realized that you do not need a robot that provides a specific force or a specific accuracy for all times. With a bit of intelligence, you can relax those needs and behave like robots like a human being. To achieve those tasks, "said McKinley.

Blue is constantly capable of catching 2 kg weight with fully expanded weapons. But unlike conventional robot designs, which are characterized by a consistent "force / current limit", "blue" is designed to be thermal-limited, "said McKinley. 

That means, like a human being, It can put a force more than 2 kilograms in a quick burst, as long as its thermal boundary does not take place and it takes time to relax or cool down. This is how it is One The person can lift the laundry chair and easily take it to a room, but it may be that the same washing plate cannot be taken without a break more than a mile.

"Essentially, we can get out of a weak robot," Gile said. "And a weak robot is just safe, the most robust robot is the most dangerous. We wanted to design the weakest robot, which can still be really useful stuff."

"The researchers were developing AI for the existing hardware and, almost three years ago, we started thinking, 'Maybe we can do something else, maybe we can think that we Which hardware can be used to enhance AI and work on those two paths. 

Together, at the same time, "McKinley said." And I think that this is a very dramatic change as much All research is done. "