A new bionic-robot hand is not only akin to a human one but also learns from experience
Scientists were able to create a bionic robot in 2013, but the best it could do was walk—when left to handling things, it ended up breaking them. While robots have since then evolved with algorithms and machine learning, hand movement has still been one of the most intricate problem for robotics. But researchers at the University of Washington have developed a robotic hand that simulates the human hand with fingers, tendons, joints and over a hundred sensors which can perform most functions that humans can—better still, it can do them faster. Moreover, the robotic hand learns from shadow hand movements and its algorithm lets it improve its dexterity.
Once trained, the robotic hand can also accomplish tasks which require in-hand manipulation—rolling, pivoting, bending, sensing friction and other things. While robots are already being used to perform surgeries, the pace has been slow. If perfected the robotic hand can make the task easier and faster. Though the cost at $300,000 is a barrier for the technology being used for commercial and industrial purposes, for now, it presents an opportunity for other scientists to create self-learning robots that simulate the human body more closely. With many researchers working on bionic arms to help amputees, this technology can prove very handy.