Nvidia is focused on making humanoid robots safer to operate in close proximity to humans. Recognizing robotics as one of the largest opportunities for humanity, Nvidia emphasizes the importance of intelligent, reliable, and safe robots as they increasingly work alongside humans. Rather than building robots themselves, Nvidia provides core technologies that enable companies to develop intelligent and dependable robots. The goal is to move beyond the current cautious behavior of robots that maintain a safe distance, enabling them to operate more closely and effectively with humans.
For robotics to truly proliferate, Nvidia identifies five key requirements: intelligence and capability, reliability, safety, cost-effectiveness, and non-creepiness. While significant progress has been made in digital AI with tools like ChatGPT, physical robots require much higher accuracy and autonomy since there often won’t be a human in the loop to intervene. This demands near-perfect precision and safety, which must be integrated from the chip level through to the software and application layers, ensuring the systems are explainable and certifiable by third parties.
Nvidia has developed specialized computing platforms, such as the Jetson, designed specifically for robotics. These platforms provide the high computational power needed for real-time processing, energy efficiency, and general-purpose programmability essential for running complex AI models on robots. Nvidia’s approach involves a “third computer” concept—besides the computers used for training and simulation, there is a dedicated onboard computer inside the robot’s “brain” to handle inference and decision-making in real time.
The biggest challenge in humanoid robotics remains developing a sufficiently general-purpose and accurate “brain” or AI system. While mechanical and hardware advancements have been impressive, the software intelligence has yet to reach the breakthrough moment seen in digital AI. Nvidia believes this moment is imminent, and as AI models improve, safety will continue to be a critical design consideration integrated throughout the entire technology stack.
To enhance safety, Nvidia is leveraging its extensive experience from autonomous vehicle development, incorporating multi-layered safety features starting at the chip level with functionally safe SoCs, redundant hardware controllers, and safety-focused operating systems. On top of this, AI models combining computer vision and large language models provide situational awareness and reasoning capabilities. These systems can be rigorously tested and certified through simulations and third-party evaluations before deployment, ensuring humanoid robots can safely and reliably operate alongside humans in real-world environments.
