9 March, 2016, USA: Google has now taken up the challenge to teach robots to grab random objects. In order to shape this research, the technological giant has made a research team which will be equipped with 14 robots on which they will be conducting their study.
The team will be focusing on various steps to resolve the issue. First, it will be working on making robots analyse their environment and then create a plan on how to grab the object and then, the final step will be execution.
Google is now using these robots to train a deep convolutional neural network (a technique that’s all the rage in machine learning right now) to help its robots predict outcome of their grasps based on the camera input and motor commands. It’s basically hand-eye coordination for robots. In a statement, the team revealed that during the initial stage of the program, researches faced tough time training these machines and it took around 3,000 hours of practice before it saw the beginnings of intelligent reactive behaviors.
“The robot observes its own gripper and corrects its motions in real time. It also exhibits interesting pre-grasp behaviors, like isolating a single object from a group. All of these behaviors emerged naturally from learning, rather than being programmed into the system,” team wrote.