KAIST Researchers Use AI to Develop a Soft, 3D-printed Robotic Hand That Can Grip an Egg, 1 Kg Water Bottle
aiinsider ·
Researchers at KAIST, in collaboration with Seoul National University of Science and Technology, have used machine learning to develop a soft, highly stretchable 3D-printing material. The elastomer can stretch to over six times its original length and is compatible with Digital Light Processing printing, overcoming a common tradeoff between printability and durability. The AI model was trained on experimental data linking material formulations to properties, and it identified a recipe that balances flow and stretchability. The team demonstrated the material by printing soft actuators and a robotic hand capable of lifting a one-kilogram water bottle and grasping fragile and rigid objects like eggs, glass bottles, and a computer mouse. According to the researchers, the work shows that combining experimental data with AI can accelerate discovery of optimal material combinations. They expect this approach to speed up development of materials for soft robotics, wearables, and custom medical devices. The study was published in Nature Communications and received support from South Korean government ministries.
aiinsider