Huge Goal – the parent company of Facebook, Instagram, WhatsApp and Messenger – announced this Friday artificial intelligence (AI) robots that can develop “challenging” sensorimotor skills thanks to two developments: Adaptive Coordination of Sensomotor Skills (ASC) and visual cortex ( VC-1).
The VC-1 is the first perceptual model that supports a wide range of sensorimotor skills, environments, and imitation.
“VC-1 was trained with videos of people performing daily tasks from the innovative Ego4D dataset created by Meta IA and academic partners. And the VC-1 meets or exceeds best-known results on 17 different sensorimotor tasks in virtual environments,” explains one post.
ASC, which performs “nearly perfect” in physical environments, also participates in challenging tasks such as mobile copying with a robot, with 98% success, the statement said. B. Go to an object, pick it up, move it to another location and place the object.
One of the hallmarks of AI is that it needs data to learn, and in both cases “new ways for robots to learn have been developed using video of human interactions with the real world and simulated interactions within simulated worlds”.
In addition, the description states that Meta is interested in artificially developing a visual cortex, a copy of the brain region that allows an organism (along with the motor cortex) to convert vision into movement.
“The module is an AI system that allows an artificial agent to convert camera inputs into actions,” the statement said.
Source: Ultimahora
Source link
Karen Clayton is a seasoned journalist and author at The Nation Update, with a focus on world news and current events. She has a background in international relations, which gives her a deep understanding of the political, economic and social factors that shape the global landscape. She writes about a wide range of topics, including conflicts, political upheavals, and economic trends, as well as humanitarian crisis and human rights issues.
