Nvidia-Backed Startup Revolutionizes Robot Training with AI-Powered Platform
Nvidia-backed startup General Robotics has developed an AI-driven platform called GRID that significantly speeds up robot training time. The company claims that its software, GRID, reduces robot onboarding by as much as 99.7%, and cuts the time to import new AI models by as much as 99.5%. Additionally, it slashes skill transfer from one form factor to another, such as from a robotic arm to a humanoid robot, by as much as 97.9%.
GRID's architecture is designed to be efficient and flexible, bringing together decades of robotics knowledge into one agent-first platform with intelligence that compounds. In some cases, the software has even trained a robot on a new skill by simply viewing one short video. The platform chooses a video-to-simulation approach, reconstructs the motion in simulation, generates its own training data, then trains and deploys the new skill.
The company's CEO, Ashish Kapoor, notes that 'people have been trying to deploy advanced robotic systems the same way for 50 years.' He believes it's time to rethink the approach with the rapid advances in AI. GRID learns and accelerates recursively, leaving artifacts behind that enable later skills to be deployed in as little as 10-15 minutes.
General Robotics has already deployed its technology with top companies such as a U.S. automotive manufacturer, a global port operator, an oil and gas producer, and multiple government agencies. The company's software is built using AI, with more than 80% of its own code agentically written.