Genesis AI, a global robotics firm, unveiled GENE-26.5, a robotic brain that gives general-purpose robots human-level hand dexterity. The system pairs a foundation AI model with a human-scale robotic hand and a novel data engine that captures motion from a glove.
Co-Founder Theo Gervet said the system learns from human hand movement data, enabling robots to perform multistep tasks like cooking a full omelet from scratch. The robotic hand mirrors human anatomy and uses micro-adjustments for grip and balance, solving a long-standing challenge in robotics.
Genesis AI’s glove-based data collection is 100 times cheaper and five times more efficient than traditional methods. Training also uses internet video and simulation to accelerate learning. The company plans initial deployments in warehouses and manufacturing, with home applications to follow.
Safety includes extensive simulation testing and adherence to regulations. Gervet said small-scale deployments with partners could begin later this year.