In the Robotics field, people are always trying to create sensors that can imitate (or get close to) how humans see, smell, hear, and feel. In this specific case, the project focuses more on improving how a robot can feel different materials using an optical sensor called GelSight. This project was conducted by MIT and Northeastern students, and it is essentially a cube with a slab of rubber on one side and different colored lights on the other sides. The rubber has a metallic coat on one side, and when something presses against this cube, the rubber conforms to that object. Due to the metallic coat, the lights will be reflected inside the cube in such a way that a camera will be able to capture the reflections and create an image of the object through a complex algorithm. What this means is that a robotic arm equipped with this sensor can pick up a USB cable hanging freely from a hook, look at the USB symbol, calculate its position relative to the USB port and plug it. This shows incredible dexterity and spacial resolution coming from a robot. In my opinion, this concept can be branched into many other areas and has the potential to offer amazing results. It is basically like a hand: we can pick things up, and without seeing, we can figure out what it is through touch. It is taking robots one step closer to humans in an innovative and interesting way.
More information and videos showing how it works can be found here: http://newsoffice.mit.edu/2014/fingertip-sensor-gives-robot-dexterity-0919
No comments:
Post a Comment