Computer scientists at Brown University have created an algorithm that gives robots handwriting and drawing capability, which they called a step toward machines that can communicate and collaborate more seamlessly with humans.
Brown's Atsunobu Kotani said, "Just by looking at a target image of a word or sketch, the robot can reproduce each stroke as one continuous action."
The algorithm employs deep learning networks that analyze word or sketch images, inferring the likely sequence of pen strokes that generated them so it could replicate those images.
When trained on a set of Japanese characters, not only could the algorithm reproduce those characters and their generative strokes with about 93% accuracy, but it could replicate dissimilar character types it had never before seen.
From Brown University (RI)
View Full Article
Abstracts Copyright © 2019 SmithBucklin, Washington, DC, USA