Search
On June 17, NVIDIA's GEAR Lab introduced an innovative robotic self-learning solution. Leveraging a 'coaching team' comprised of AI programming agents, this system devises training regimens for robotic arms with minimal human oversight. This enables the robots to acquire skills in cutting plastic cable ties, organizing components, and precisely inserting GPUs into motherboard expansion slots. The foundation of this solution lies in the ENPIRE framework, which endows eight Codex agents with robotic capabilities, GPU resources, and defined task objectives, facilitating autonomous agent management. The robots are capable of independently executing high-precision tasks, such as securing cable ties, arranging nails, and installing graphics cards. They can even operate overnight to refine their skills, with researchers only required to review the subsequent training reports.
-
C114 Communication Network -
Communication Home
