In fractions of a second before capturing an object, your brain calculates all the movements required to safely reach and capture the object. This is a subconscious approach that is the result of years of development and learning in childhood and is now being used by robotics researchers for their own creations. Not only is Festo's new BionicSoftHand remarkably skilled, it uses the AI to find out how to properly manipulate and manipulate an object before making actual moves. The robots that move heavy parts in a factory are made of steel and pneumatic components that make them strong and fast, but not very forgiving. If a towering industrial robot from FANUC accidentally came in contact with a human while performing its movements, it would cause serious injury. In comparison, soft robots are constructed with flexible materials such as smart fabrics and inflatable bladders. As a result, they have a commandment and a compassion. So, if you have contact with a human being while working, they will not cause immediate injury or damage to yourself.
For maximum safety, the new BionicSoftHand from Festo does not have a rigid skeletal structure. Instead, it has a series of inflatable bellows surrounded by a fabric sheath made of elastic fibers that move and flex with the movements of the hand as air is pumped to create movement. It works much like the muscle and tendon system of the human hand, but the soft components mean that the direct interaction with humans is much safer.
The digits of the BionicSoftHand are also equipped with inertial (motion) and force sensors that facilitate the passage through the use of flexible printed circuit boards and wiring that will not latch when deformed. These sensors provide feedback to the robot control systems as to when the hand and digits move, as well as when they are stopped. This indicates that contact was made with an object or that an object is in a position where it is not possible to move on. They essentially provide a sense of touch, which is important given the other capabilities of this robot hand.
Robot arms and manipulators in the factory settings are programmed to repeat very specific movements endlessly. You may occasionally notice slight variations in the tasks you perform, but most of the time they pick up exactly the same object that is always in one particular place in your hand and move it to another predefined location.  Like humans, BionicSoftHand learns to capture and manipulate objects and perform tasks themselves. However, this happens through virtual testing, rather than risking an accident in real life. As a toddler, you have probably dropped a cup of milk 100 times before you master it, grabbing a cup and taking a sip; However, the BionicSoftHand can make these mistakes billionfold with a digital twin powered by AI and machine learning, without spilling a single drop of milk in everyday life.
The robot has a specific goal, eg. For example, repositioning a 12-sided shape in the hand so that a certain corner points upwards, but never said how to move the numbers to manipulate the object. A depth-detection camera creates a digital duplicate of a real object that the hand must interact with, allowing countless virtual hands to manipulate the copy of the object until a solution is found. Then the real BionicSoftHand takes over. It is able to learn much faster than a toddler can, and it is better to use the movements and techniques that it has already found for new challenges.
At this first look at the technology that Festo has provided, a single BionicSoftHand is shown moving a 12 skillfully cube around, but it does not take much imagination to imagine a pair of these hands learning fast to solve a Rubik's Cube: a skill I do not have to master yet. It costs millions of dollars to design, develop, and program an industrial robot to perform a single job in a factory, but this self-learning approach means that only one robot can help with countless tasks throughout the house.