MIT's robot can find your lost stuff

MIT has developed a new robot that is designed to help users find misplaced items. The robot is called RFusion, and it's comprised of a robotic arm fitted with a camera and radiofrequency antenna attached to its gripping mechanism. The design combines signals from the RF antenna with visual input from the camera attached to the gripping mechanism.

With all the hardware and software combined, the robot can utilize visual input to locate and retrieve an item, even if that item is underneath other objects and completely out of view. The prototype robot utilizes RFID tags attached to items that send a signal reflected by its antenna.

Radiofrequency signals are perfect for this type of system because the signals travel through most surfaces. MIT's robotic arm leverages machine learning allowing it automatically zoom in on the exact location of the object it's searching for. It's able to move directly to the top of the object, grasp it, and verify that what it picked up is the correct item.

MIT fully integrated the camera, antenna, robotic arm, and AI allowing RFusion to operate in any environment without requiring a specialized setup. Finding lost objects in the home might prove useful, but one of the robot's biggest potential areas of use is for tasks like sorting through merchandise to fill orders in a warehouse.

Project researchers have stated that the prototype isn't fast enough to be used in industry or the home just yet. Researchers do believe their robot has applications in manufacturing and warehouse environments. The robot begins its search for a specific item by searching for the object using its antenna. The antenna bounces signals off the RFID tag to identify the area where it's located, and the camera then narrows down the object location. RFusion has a 96 percent success rate for retrieving objects fully hidden under a pile.