“In order for a robot hand to recognize an object and grasp it with appropriate force, the AI that serves as its “brain” needs to be trained with a large amount of data. However, it is difficult to actually prepare objects of various sizes and shapes for training. Ascent’s generative AI technology generates 3D “teaching data” to train AI in a virtual space.” (Excerpt from the article)
More details about the article https://www.nikkei.com/article