Dog training technique also works on robots



Using the reward technique, which has been shown to be highly effective in dog training, U.S. researchers have taught a robotic arm a series of new tricks, such as building brick towers. learn to perform complex tasks such as washing dishes and working on production lines. The important thing in the new experiment is that the robotic arm initially had no idea what bricks were and how to handle them. He learned on his own what he had to do thanks to the technique of rewarding, or “positive reinforcement” as behavioral psychologists call it. “He never gives up and keeps trying to build the bricks until he achieves the goal in 100% of cases,” said Andrew Hood, a Johns Hopkins University PhD student in Connecticut. I know the reward works. “This was the inspiration for the design of the learning algorithm,” he says. His team’s publication, “Good Robot!” in dogs. Where a dog would win a cookie, the robot would be rewarded with floating points in a kind of “point system”. The robotic arm, named Spot, first began experimenting with cubes, not knowing exactly what to do. does. He soon learned that the moves leading to stacking the cubes offered reward points, while the wrong moves offered him nothing. “What the robot wants is the highest possible score. “He quickly learns the right behaviors to get the best reward,” explains Hood. He now hopes that the positive amplification technique will help develop robots that learn complex tasks, such as assembling components on production lines or autonomous driving. , such as assembly and care for the elderly. “Today we do not know how to plan such work. But our work shows that there is a future in the idea that robots can learn to achieve such goals in real conditions in a safe and efficient way. ā€¯Follow it on Google News and be the first to know all the news See all the latest news from Greece World, in



Source link

Leave a Reply

Your email address will not be published. Required fields are marked *