RLWRLD's RLDX-1 Enables Humanoid Robot Dexterity
A humanoid robot demonstrated sorting black and white socks at RLWRLD's "Dexterity Night" event in San Francisco, according to Interesting Engineering. The demo used onboard cameras and five-fingered hands to detect colors, grasp items on a moving conveyor belt, and place black socks into a separate container, per reporting from The Chosun Daily and Interesting Engineering. RLWRLD unveiled a dexterity-first foundation model called RLDX-1 that integrates vision, motion, memory, and torque streams, per coverage in The Robot Report and a technical report posted on arXiv. RLWRLD has also captured human hand motion and workplace footage with body cameras for training data, the company describes on its website and in media coverage. Editorial analysis: This combination of multi-stream models, long-horizon memory tokens, and real-world human data represents a notable step toward practical robotic manipulation for industrial and service tasks.
