Robotics has been pivotal in Nvidia’s impressive growth trajectory over recent years. When the chip manufacturer introduced GR00T in March last year, it was hailed as a groundbreaking milestone for humanoid robotics. Many leading names in the field, including 1X Technologies, Agility Robotics, Apptronik, Boston Dynamics, Figure AI, Fourier Intelligence, Sanctuary AI, and Unitree Robotics, were involved from the outset.
During CES on Monday, Nvidia CEO Jensen Huang unveiled another crucial component of GR00T: Blueprint. This new feature focuses on imitation learning, a key technique for teaching robots new skills. This approach is especially significant for humanoid robots, whose designs are modeled after the human instructors guiding them.
Imitation learning involves a person demonstrating an action, which the robot then replicates. It’s a highly effective way to train robots for automating tasks, particularly in industries like manufacturing and warehousing, where humanoid robots are beginning to make their debut.
Teleoperation also plays an integral role in this process, enabling robots to be taught remotely by digitizing human actions in a simulated environment.
With GR00T Blueprint, users can create these digital actions using Apple’s Vision Pro. The movements are captured as a digital twin, which the robot can then simulate and repeatedly execute, further refining its performance.
