Over the past two years, researchers have taught robots to play with toys to AI Lab, which is operated by Autodesk, a manufacturer of AutoCAD and other 3D design software.
Teachings may not be correct words. Specifically, we mounted the necessary tools to learn how to assemble a Lego block using the same learning method as a child, for a robot nicknamed Brickbot.
Although it can be said that it is simple, the task of enabling robots to learn as human beings is huge even in small formats.
A Autodesk spokesperson said, "It's a complicated task." "This project uses sensor data and machine learning to estimate what is happening in the environment of the robot and adapts it to accomplish that task on the spot.
Robots are suitable for following strict protocols, but this limits their usefulness. Currently, the programming robot for industrial robots requires a lot of work. The new generation of collaborative robots has made it easy for many light industrial applications, but working with robots on a single line is an area of experts.
By giving the robot its own learning function, it is possible to promote the spread of automation by reducing the entry cost while securing a new level of productivity.
Of course, in such a project, all sorts of questions about the extent to which machine learning of robot application progresses before people feel a bit uncomfortable will be issued.
This project began with two industrial robotic arms, with researchers adding various types of cameras and sensors.
A neural network, which is a moderately modeled computer system of the human brain, enables a robot to process intelligently and to form productive response behaviors from its environment when tasks are assigned There.
"From the plastic bricks, we were able to process the project freely, from design to completion of the finished product," Yotto said. Software architect with a doctorate in robotics engineering, Koga. "We are going to go to the next step now and we plan to work closely with customers in manufacturing and construction customers to see how Brickbot technology can be applied to the real world.
In the real world, we will not test the team's performance when creating robotic systems that can learn productivity. Can such a system reassemble the components with electron beam after self-induction test or day of error?
The adaptability and context awareness required in such a scenario is far more important than what Brick Bot has achieved so far. However, it is closer to the day that robots can learn to execute complicated tasks in less time and less error than when human programmed.
In other words, robot programmers may become different jobs at the beginning of automation.