LAS VEGAS : At the global technology exhibition in Las Vegas, Nvidia announced an integrated package of core models and technologies aimed at unifying global standards for developing intelligent robots. The move significantly strengthens the company’s influence over how machines perceive, reason and act in the physical world.

A major highlight of the announcement was the introduction of advanced Cosmos models, designed to help robots understand physical laws and predict outcomes before executing actions. Among them, Cosmos Reason and Cosmos Predict enable machines to perform complex visual and language-based reasoning, allowing robots to adapt to dynamic environments with near human-like flexibility.

Nvidia also introduced Isaac GR00T, a visual-motor language (VLA) model that serves as the cognitive brain of robots by translating software code into precise physical movements. Paired with the GR00T-Mimic tool, developers can now train robots using simple human demonstration data, a breakthrough that could compress years of robotics research and development into a much shorter timeframe.

Further expanding its robotics ecosystem, the company unveiled the Jetson T4000, a next-generation embedded computing platform built to operate directly inside robots. The ultra-powerful processor delivers high-speed decision-making with exceptional energy efficiency, enabling robots to act in fractions of a second without relying on constant cloud connectivity.

With these launches, Nvidia is working towards what it describes as a unified development environment, effectively a “real operating system for physical intelligence.” The goal is to make building robotics applications as intuitive and scalable as developing smartphone apps, accelerating global innovation in robotics.