Nvidia seeks to make humanoid AI robots safer around humans - The Mercury News

NVIDIA is advancing humanoid robot technology to improve safety in human environments. As AI and robotics converge, humanoid robots are transitioning from lab settings to real-world applications, part... Read original →

NVIDIA’s adaptation of its autonomous driving perception stack Halos to humanoid robots leverages its 3nm AI chip and EUV process leadership to extend computational dominance into the physical world. This move pressures upstream sensor suppliers to accelerate low-latency, high-precision mmWave-vision fusion modules and catalyzes ROS 2.0 safety middleware standardization. Regulatory-wise, pending EU/US functional safety frameworks akin to ISO 13849 could add 6–12 months to certification timelines, raising barriers for smaller entrants. Tesla’s Optimus and Agility’s Digit already boast closed-loop validation, while Meta and Amazon may bypass IGX Thor via in-house AI silicon. Over the next 18 months, industrial settings will serve as real-world stress tests for safety algorithms; home deployment remains bottlenecked by liability ambiguity—precisely why NVIDIA launched a dedicated lab: to shape standards, not just sell chips.