If you look at the human hand for a few seconds, you will notice something impressive, it can manipulate practically any object in the world, despite all the advances in artificial intelligence, many robos are still incredibly clumsy when they need to manipulate real-world objects. open a box, hold a tool or simply pick up something delicate without breaking it, but now, a North American startup called Origami Robotics believes it has found a way to solve that problem.

The company developed a system that combines a highly articulated robotic hand with a smart glove worn by humans. This glove precisely captures all the movements of an operator's fingers and those movements are transferred directly to the robotic hand, creating what engineers call a digital twin of the human hand.

The above allows robos to learn tasks by observing and exactly replicating the movements of a human. According to researchers, this type of approach solves one of the biggest obstacles in robotics, training data. Training robos using videos of humans usually generates imprecise information. Virtual simulations also often fail when faced with the real world, but when the human hand and the robotic hand have practically the same mechanical structure, the data captured becomes much more precise.

The ultimate goal of Origami Robotics is to create a model capable of universal object manipulation, that is, robots that can interact with virtually any object that a human could manipulate. The company's first devices are already being sent to large research and artificial intelligence laboratories, including Amazon AI laboratories interested in using this technology to train more skillful robots.