A humanoid robot developed by a South Korean startup will be used as a teaching tool in Nvidia’s education business, marking the first time a domestically built humanoid has served that role in an Nvidia training program.
Nvidia will launch a new in-person robotics education program in Seoul in October, according to the information technology industry on the 24th. The six-month program will teach college students and job seekers how to develop artificial intelligence models for humanoid robots. Korean robotics startup XYZ is taking part. It is the first time Nvidia has organized a robotics education program together with a Korean robotics company.
In the program, XYZ’s semi-humanoid robot “Deuce” will be used as a hands-on teaching tool. Deuce is a semi-humanoid robot designed by XYZ, with its physical version unveiled last month. It has two arms on its upper body and wheels on its lower body, with 32 degrees of freedom, a unit that counts a robot’s drive axes. The product is positioned as a general-purpose robot for use in retail stores, offices and homes.
The two companies’ education program includes Nvidia’s Deep Learning Institute (DLI) curriculum. DLI is an Nvidia education program that teaches the basics of GPU-based deep learning, how to use Nvidia’s software development kit (SDK) and how to apply robot development to specific industries. As the lead organizer, Nvidia will recruit instructors according to the DLI curriculum and provide programs such as mentoring by working engineers. The instruction Nvidia will handle covers how to use graphics processing unit (GPU) development tools, developing AI models specialized for humanoid robots, and digital twin simulation. Nvidia’s “Jetson” series of edge (on-site deployment) AI chips, its “Isaac Sim” robot simulation program and its “Isaac Lab” reinforcement learning platform will be the main topics covered in the training.
XYZ will take part in the overall planning and operation of the education program. In particular, XYZ will fully lead the hands-on courses that use Deuce. The goal is to implement in real environments the robot learning developed and conducted in virtual environments during the earlier part of the training. Toward the end of the program, Nvidia and XYZ will evaluate the work created by the students.
