Researchers at Google's DeepMind have made significant advances in robot dexterity, enabling robots to learn complex tasks that require precise movement. Two new AI systems, ALOHA Unleashed and DemoStart, are paving the way for robots to perform a wide variety of helpful tasks. ALOHA Unleashed helps robots learn to perform novel two-armed manipulation tasks, such as tying shoelaces and repairing another robot. DemoStart uses simulations to improve real-world performance on a multi-fingered robotic hand, achieving a success rate of over 98% on various tasks in simulation. The systems were developed using MuJoCo, an open-source physics simulator, and tested on a three-fingered robotic hand called DEX-EE, developed in collaboration with Shadow Robot. These advances bring us closer to a future where AI robots can assist people with daily tasks at home and in the workplace.
↧