A Robot Slave to Do All Your Menial Tasks

Who doesn’t want a robot to clean the house, put the dishes back in the cupboard, feed the pets and clear a cluttered table? Those types of life sucking chores are the exact reason we always dream of having the little guys around in the first place.
The problem with robots is that we are always comparing them to us. Our brains calculate at such a blazing speed that robots just don’t have the time to catch up. The easy part is observing all of the visual and auditory data. The analysis is the part where our poor number crunching minions have their troubles.
Until now, robots needed to use preprogrammed three-dimensional objects. If it didn’t recognize an object it had trouble interacting with it. A perception system is the key. The STAIR (Stanford Artificial Intelligence Robot) by Andrew Ng and Aaron Edsinger aims to perceive the 3D world in real time.
They rigged a system together with fragmented technologies: machine learning, computer vision, speech recondition and grasping hardware. The real feat, however, is the algorithm for grasping. It identifies a midpoint of graspable edges on any object, such as a handle. This may not sound like rocket science, but it makes the difference between a smart robot and a preprogrammed machine.
It still is a long way off from performing tasks, such as making me bacon and eggs in the morning. But then again, we can’t all be perfect. That is, until we’re perfected.













