Here is an innovative robotic learning technology that enables robots to imagine the future of their actions to figure out how to manipulate objects they have never seen before. UC Berkeley researchers have developed this visual foresight technology to enable robots to predict what their camera will see if they perform a task.
The robot can learn how to perform tasks without any help from humans. It builds a predictive model of the world and uses it to manipulate new objects. As the researchers explain:
At the core of this system is a deep learning technology based on convolutional recurrent video prediction, or dynamic neural advection (DNA). DNA-based models predict how pixels in an image will move from one frame to the next based on the robot’s actions.
You can read all about the research here.