NVIDIA Leads the Way in Humanoid Robotics: Introducing HOVER and Project GR00T

NVIDIA Leads the Way in Humanoid Robotics: Introducing HOVER and Project GR00T

NVIDIA has been making waves in the field of robotics, and its latest developments in AI-driven humanoid robotics are no exception. This year, the company introduced HOVER, a 1.5-million-parameter neural network specifically designed to enable advanced locomotion and manipulation in humanoid robots. Unlike conventional models, HOVER is lightweight yet powerful, allowing robots to perform intricate movements and respond to high-level commands with impressive accuracy.


Training HOVER took place within NVIDIA’s Isaac simulation suite, a GPU-accelerated platform capable of physics simulations up to 10,000 times faster than real-world speed. This powerful environment allowed HOVER to undergo a year’s worth of training in just 50 minutes on a single GPU—a remarkable feat that enables efficient model deployment without extensive real-world adjustments. This fast-tracked simulation approach helps researchers create adaptable robots ready for dynamic real-world interactions.


HOVER's adaptability is one of its defining features, supporting a broad range of input sources. It can interpret commands from various devices, including XR platforms like the Apple Vision Pro, motion capture systems, and joysticks, making it an exceptionally versatile tool for teleoperation and real-time adjustments. This input flexibility lets HOVER operate as a unified control interface for humanoid robotics, contributing valuable data for continual learning and refinement.


This advance is part of a broader commitment by NVIDIA to expand the horizons of humanoid robotics. Earlier this year, NVIDIA also introduced Project GR00T, a foundational model for general-purpose humanoid robots. GR00T enables robots to understand natural language and replicate human movements, facilitating nuanced interaction and coordination. By observing human actions, GR00T-powered robots learn dexterity and adaptability essential for real-world functionality.


Together, HOVER and GR00T exemplify NVIDIA’s dedication to creating intelligent, adaptable robots that can seamlessly integrate with human environments across sectors—from healthcare to customer service. These cutting-edge technologies underscore NVIDIA’s leadership in making humanoid robotics a practical, transformative technology that brings robots closer to functioning as collaborative companions in our everyday world.

Zack Kamal

investor at Link prospects

1 周

It's inspiring to see NVIDIA pushing the boundaries of AI in robotics. Your insights into these groundbreaking innovations are truly eye-opening, Aamir!

回复
Merab Asher

Technical Recruiter | HR Manager | Compliance & Policy Expert | Employee Relations Strategist | Culture Innovator | Event Manager

2 周

Interesting

回复
André Pereira

Business Growth Architect | Partnership Builder | Strategic Thinker

2 周

NVIDIA's HOVER sounds revolutionary! How do you see its neural network impacting the development of humanoid robotics in various industries? ?? On another subject, it would be great to connect! Feel free to send me a request.

回复
Hafiza Shamza Hanif (AI Developer)

AI Developer || LLM || RAG || Data Scientist || ML || DL || NLP || Chatbots || Researcher|| Actively seeking Remote job in the field of AI

2 周

Very informative

回复

要查看或添加评论,请登录

Aamir Ahmed的更多文章