Jul-03-18
Researchers have developed an artificial neural network that allows robots to grasp a particular object in a cluttered and changing environment.
Developed by a team from Queensland University of Technology, the artificial neural network was applied in conjunction with a depth-mapping camera, allowing a two-fingered robot to scan the environment and match each pixel to its ‘grasp quality.’ The system enables the robot to map the environment in a single pass, saving time, and delivered up to an 88 percent grasp success in a changing environment, and 92 percent in a static environment.
Somebody has to implement these robots in those machines where you insert a coin to get a change to grab an item. I guess the accuracy could improve around 100 percentage points. *Feeling funny* Posted by Anon ymous on July 4, 2018
Show 1 Comment