NVIDIA Establishes the Most Well-funded AI Lab Specializing in Embodied Intelligence Research
-
NVIDIA recently announced the establishment of a General Embodied Intelligence Research Group, dedicated to developing intelligent systems with the ability to understand, reason, and interact with the physical world. The research team is co-led by NVIDIA senior scientist Jim Fan and his colleague Yuke Zhu, who believe that future mobile robots will become autonomous entities, as ubiquitous as iPhones.
They are working on creating foundational agents—AI with general capabilities that can learn action skills in various virtual and real-world scenarios. Jim Fan stated that with NVIDIA's soaring market value, they will have sufficient funding to support research and development in robotics, gaming AI, and simulation, making it one of the world's most well-funded embodied intelligence labs. NVIDIA's previous research achievements in the field of embodied intelligence were prominently showcased. This includes the Eureka project, which uses GPT-4 to generate reward functions, trains robots to complete multiple complex tasks, and achieves training speeds 1000 times faster than real-time.
The Voyager project applies GPT-4 to Minecraft, rapidly advancing the technology tree and performing reasoning and operations through game visuals, becoming the first intelligent agent capable of skillfully playing Minecraft. Additionally, the MineDojo project, which won the NeurIPS2022 Best Paper award, proposed an "embodied GPT-3" composed of multiple agents and transformed Minecraft into an AGI research playground. The industry has high expectations for the general embodied AI research team established under the leadership of Jim Fan. Since last year, research on embodied intelligence has gradually gained attention, including projects like Stanford University's household robot initiative. This highlights the urgent demand for intelligent systems capable of understanding, reasoning, and interacting with the physical world. As a result, there is great anticipation for NVIDIA's future research outcomes.
NVIDIA's formation of the general embodied AI research team aligns with the future trends of intelligent system development and represents a significant investment in the field of artificial intelligence. Their previous achievements in embodied AI research have already drawn widespread attention, and their future progress is highly anticipated.