DeepMind Technologies is an artificial intelligence unit owned by the company “Alphabet.” They are the leaders in Deep Reinforcement Learning technique by which neural networks can learn from experience. With this technique's help, Deep Mind has developed AI, which can beat human masters at games like Go and Chess; also, it can become a top player at a computer game called Starcraft.
Using a similar technique, a researcher has recently developed a robot dog that can cope with situations it has never seen before, like stairs or slippery surfaces. This unknown applicability of Artificial Intelligence in robotics is a tremendous development in Robotics Market.
The researcher taught the robot to do things by taking up the method used to teach young children how to play football. Initially they will be taught stuff like shooting, passing, and dribbling. When they have mastered all the basic skills, they are asked to play simple matches to test their skills and how to put all that they have learned together. It is the most natural way humans learn anything, and researchers have been brainstorming on how to use this method for robots for some time now.
In this research, the robotic dog was also taught only two skills initially that were trotting, walking and fall recovery. Each of these skills was developed in a specific artificial neural network. These networks depend on layers of millions of tiny connections between clusters of mathematical computations and nodes and can modify themselves as per their training.
The two essential skills of trotting, walking, and fall recovery were treated as the foundation with which the other six neutral networks were created. The researcher then made an AI that would combine all their skills to solve some problems like getting up from various positions and walking over to a target. The robot's best thing is to quickly adjust to entirely new scenarios and continue towards what it was doing.
Even though this new robot can be considered a significant feat in the industry, eventually, it would be essential for these robots to be beneficial for an individual. This can be achieved with the addition of features like robotic hands, vision systems, hearing ability, etc., in turn bringing a high level of complexity.
In the coming future, robots will come to be used by humans much more as they start to perform small tasks. They will be more interactive, creating a safe environment and increasing their usefulness a little more.