The day when robots can cook dinner, clear the kitchen table or empty the dishwasher is still a long way off. But a research team of researchers at the University of Texas made significant progress with their new robotic system that uses artificial intelligence to help robots better recognize and remember objects.
“If you ask the robot to pick up a mug or bring a bottle of water, the robot needs to recognize these objects,” noted a member of the research team.
The new technology is designed to help robots identify a variety of objects found in environments such as homes and distinguish similar versions from common objects such as water bottles of different brands, shapes or sizes.
Different packages of artificial food are used to train Ramp, a robot in a university lab. Ramp is a mobile manipulator robot from Fetch Robotics that stands on a circular mobile platform and is approximately 4 feet tall. He has a long, seven-jointed mechanical arm, ending in a square hand with two fingers for grasping objects.
According to researchers, robots learn to recognize objects in a similar way to how children learn to interact with toys.
“After pushing an object, the robot learns to recognize it,” said a member of the research team. “With this data, we train an AI model so that the next time the robot sees an object, it doesn’t have to push it again. The second time it sees an object, it just picks it up.”
What’s new about the researchers’ method is that the robot pushes each object 15 to 20 times, while previous interactive perception methods use just one push. Multiple pushes allow the robot to take more pictures with its RGB-D camera, which includes a depth sensor to explore each item in more detail. This reduces the chance of errors.
Recognizing, distinguishing and memorizing objects, called segmentation, is one of the main functions that robots need to perform tasks.
Source: Science Daily
