Festo is widely known among amateurs and fans of modern robotics – in particular, among those interested in hybrid systems of robotics. A few years ago, the company presented an unusual pneumatic robot BionicKangaroo, and now we are pleased to present a new iteration of the old system, whose name this time is BionicSoftHand. It is worth noting that, despite its previous experience of such an assembly, the new robotic system is a living experiment of Festo, since the mechanical arm based on pneumatics this time uses a complete set of self-learning algorithms.
In other words, BionicSoftHand is not just a robot-pneumatic hand, suitable only for a strictly defined and specified set of movements and manipulations, but a full-fledged system of constantly developing artificial intelligence, which uses deep analytical algorithms for its self-study. Experts note that the pneumatic arm is actively tested and trained using the so-called cubic test – the hand is invited to interact with a 12-sided cube, each side of which is marked with a certain color.
Giving a command to the robot to raise one or another color, specialists allow the machine to independently learn from its mistakes and approaches to work, thereby avoiding the clichéd type of training. Moreover, what distinguishes this system from many other third-party ones is the fact that the robot uses a system of connected training — in which, having studied one skill and an aspect, one unit transmits this knowledge and skills to everyone else on the same network as it can be extremely convenient and promising.
It is worth noting that Festo strives to improve its BionicSoftHand robotic pneumatic arm to such an extent that the artificial intelligence algorithms contained in it are constantly improved on their own, thereby avoiding the initially limited and defined scope of such training. What can have far-reaching consequences in the key of the company's developments.