Meta is diving deeper into AI research with a focus on enhancing its metaverse ambitions. The company recently introduced Meta Motivo, an AI model designed to control the movements of digital avatars, potentially revolutionizing the metaverse experience.
Motivo: A Behavioral Foundation Model
Meta claims that Motivo is the first of its kind, a behavioral foundation model that enables virtual avatars to perform complex whole-body tasks, making their movements and interactions in the metaverse more realistic and seamless.
Unsupervised Reinforcement Learning
Through unsupervised reinforcement learning, Motivo can adapt to complex environments and perform a wide range of tasks. A novel algorithm and an unlabeled dataset of motions allow Motivo to learn human-like behaviors while retaining the ability to infer actions without specific training.
Enhanced Avatar Movement and Interaction
A video demo showcased Motivo’s capabilities, showing a humanoid avatar performing dance moves and kicks using whole-body movements. Meta aims to create more realistic and human-like avatars through this technology.
Applications in the Metaverse
Motivo can reportedly solve various whole-body control tasks, including motion tracking, reaching specific poses, and optimizing movements based on rewards, all without requiring additional training.
Reality Labs and Metaverse Development
Meta’s Reality Labs division, responsible for metaverse initiatives, has been exploring various technologies to enhance the metaverse experience. This includes Hyperscape, a tool that transforms smartphone camera feeds into photorealistic metaverse backgrounds.
Meta’s continued investment in AI and metaverse technologies, despite financial challenges, demonstrates its commitment to shaping the future of virtual interactions and experiences.