MIT robot senses hidden objects using penetrative radio frequency

Satsuki Then - Apr 1, 2021, 7:39am CDT
MIT robot senses hidden objects using penetrative radio frequency

Researchers worldwide have been working hard to give robots the ability to better sense the environment around them. Before robots can work side-by-side with humans, they have to be able to sense humans and other objects for safety and to be useful. Researchers from MIT have developed a new robot able to sense objects, even if hidden from view, using radio waves.

The radio waves used by the robot can pass through walls sensing hidden objects. The robot is called RF-Grasp, and it combines the new powerful sensing technique with traditional computer vision to locate and grasp items that might be blocked from view. Scientists on the project believe the advance might lead to streamlined e-commerce fulfillment and warehouses or improved ability to help machines choose specific items, like a screwdriver from a box full of tools.

The robot could prove particularly useful as e-commerce continues to grow. The challenge for robots used to pick up items for e-commerce orders today in warehouses worldwide is that the robots struggled to locate and grasp objects in a crowded environment. Researchers say that perception and picking are two major roadblocks in the industry today.

When relying on optical vision alone, robots cannot perceive an object packed away in a box or hidden behind another object on a shelf. While light waves can’t pass through walls or other objects, radio waves can. MIT’s RF Grasp robot uses a camera and RF reader to find and grab tagged objects even when blocked from the camera view.

The robot is simple and is just a robotic arm attached to a grasping hand. The RF reader is independent from the robot and relays tracking information to the robot control algorithm. Researchers say integrating RF tracking and a visual picture of its surroundings was difficult because the robot had to learn to choose which data stream is more important at each point in time. The team believes the robot could one day perform fulfillment in busy e-commerce warehouses and remove some of the industry’s limitations.


Must Read Bits & Bytes