[ad_1]
In a fraction of a second before you can grasp an object, your brain calculates beforehand all the movements needed to reach and seize it safely. It is a subconscious approach that results from the years of development and childhood learning that robotic researchers are now using for their own creations. Festo's new BionicSoftHand is not only remarkably clever, but uses artificial intelligence to understand how to properly hold and manipulate an object before it makes a real move.
BionicSoftHand is another creation that takes a "soft" approach to robotics. The robots you see moving heavy parts into a plant are made of steel and pneumatic components, which makes them robust and fast, without forgiving too much. If an imposing industrial robot, such as that of FANUC, accidentally came into contact with a human during the execution of its movements, serious injuries would result. In comparison, soft robots are built with soft materials such as smart fabrics and inflatable bladders. As a result, they have gift and conformity. So if they meet a human while working, they will not hurt themselves immediately.
For maximum safety, Festo's new BionicSoftHand does not have a rigid skeletal structure on the inside. Instead, it has a series of inflatable gussets surrounded by a knitted skin fabric made of elastic fibers that move and flex with hand movements when air is injected to create motion. Its functioning is similar to that of the muscular and tendon system used by the human hand, but the soft components mean that it is much safer for the man to interact directly with him.
The BionicSoftHand figures are also equipped with force and inertia sensors, facilitated by the use of flexible printed circuit boards and cabling that will not break when deformed. These sensors provide the robot's control systems with information on when the hand and the numbers move, but also when they stopped, indicating that a contact has been made with an object or an object. is in a position where he can not be. further moved. They essentially provide a sense of touch, which is important given the other capabilities of this robot hand.
Robot arms and manipulators in a factory setting are preprogrammed to continually repeat very specific movements. They can sometimes account for very small variations in the tasks they perform, but mainly, they recover the same object that will always be at a very specific place in the hand and move it to another preset location.
Like humans, BionicSoftHand learns to grab and manipulate objects and to perform tasks on its own. But this is done by virtual trial and error, instead of risking an accident in real life. As a toddler, you have probably dropped a cup of milk 100 times before mastering how to grab a cup and take a sip; but the BionicSoftHand can commit these mistakes billions of times using an AI-optimized digital twin and machine learning – without ever pouring a single drop of milk into real life.
A specific goal is assigned to the robot, such as repositioning a 12-sided shape in his hand so that a specific corner points up, but he never learned to move his fingers to manipulate that object. A depth-sensing camera creates a digital copy of a real-world object with which the hand must interact, allowing countless virtual hands to manipulate the copy of the object until it reaches the point of view. to discover a solution. The real BionicSoftHand takes over. He is able to learn much faster than a toddler and is easier to use the movements and techniques he has already understood for new challenges.
In this first glimpse of the technology provided by Festo, a single BionicSoftHand is shown that skillfully moves a 12-sided cube, but it does not take much imagination to imagine that a pair of these hands will quickly learn to solve a Rubik's cube: a skill I have not mastered yet. The design, development and programming of an industrial robot to perform a single task in a factory cost millions of dollars. However, this self-learning approach means that a single robot can help you in many household tasks.
[ad_2]
Source link