This robot can solve Rubik's cube with just one hand
After beating professional DOTA2 players, Elon Musk-backed OpenAI is using its artificial intelligence software for developing self-learning, general-purpose machines. As part of this effort, the AI research organization has developed a robotic system, which has learned the ability to solve Rubik's cube with a single hand. Yes, you read that right. Here's all you need to know about the mind-blowing capability.
OpenAI's Dactyl learns from virtual simulations
First revealed last year, the robotic system, dubbed Dactyl, uses OpenAI's artificial intelligence software to learn a task from virtual simulations. It learns from the data and applies the training to tackle real-world challenges. Last time, we saw how the robot was able to interact with unknown objects, and now, it is using the same technique for solving the Rubik's cube with one hand.
Solving the cube in just around 4 minutes
In a video shared recently by OpenAI, Dactyl, which appears like a human hand, is seen manipulating the cube with its fingers. It fumbled while handling the cube but continued to execute clumsy yet accurate moves to complete the cube in a little more than 4 minutes. Now, that's something you don't see every day, particularly from an AI-powered robot.
How Dactyl figured the way to solve the cube
Dactyl learned the way to solve Rubik's cube by going through simulations that a human would take some 10,000 years to complete. Then, after learning, it analyzed the puzzle, figured the way to solve it, and executed the movements required to do that one after the other. It knew what to do but appeared clumsy owing to the robotic nature of fingers.
This is a major step forward for AI
While the ability to solve a Rubik's cube doesn't look like a big deal, it's a major leap for OpenAI. Essentially, being successful here means OpenAI's software can be used to develop dexterous robots that can perform a range of tasks. Plus, it shows that the bot can adjust its movements to deal with unexpected hindrances in its path of action.