MIT has created a tactile system for soft robots: they can even assemble chips. This has been made possible with the help of new latex “bubbles” and proprioception systems. You can read about the development results on the university website.
The main problem in creating soft robots is that they must understand what exactly they capture and how much force they must use for this. One of the development teams solved this problem, based on previous research by the Massachusetts Institute of Technology and Harvard University, in which researchers developed a soft, cone-shaped robotic grip that shrinks even on objects like the Venus flytrap and can lift objects 100 times its weight. The developers improved this “magic ball capture” by adding sensors that let it pick up objects as thin as potato chips and classify them so that the invader can recognize them in the future.
The team also added tactile sensors made of latex “bubbles” connected to pressure sensors. The algorithm uses feedback to allow the invader to determine what effort to use. So far, the team has checked the capture sensors on objects: from heavy bottles to jars, apples, a toothbrush and a bag of cookies.
“We hope that these robots will allow the use of a new method of soft sensing, which can be applied to a wide range of different actions in the production environment: from packaging to lifting,” said Josie Hughes, lead author of a new article on the work.
The second group of researchers from the Massachusetts Institute of Technology created a soft robotic finger called GelFlex, which uses built-in cameras and deep learning to create tactile sensations and proprioception. The grip is like a person holding a cup with two fingers. Each finger has one camera near the tip of the finger, and the other in the middle. The cameras monitor the state of the anterior and lateral surfaces of the finger, and the neural network uses the information from the cameras for feedback. This allows the invader to pick up objects of various shapes.
“Our soft finger can provide high accuracy of proprioception, accurately identify captured objects, and also withstand significant effects without harming the interacting environment and ourselves”, said one of the authors of the development.