MIT Researchers Present ‘RFusion’: A Robot That Finds Lost Items Using AI

Source: https://news.mit.edu/2021/robot-finds-items-camera-antenna-1005

The idea of finding lost items in this chaotic world has been a constant problem over the years. How frustrating is it for a busy commuter to sift items to find one small object they misplaced?

Researchers at MIT unveil a robot that can resolve this issue and prove helpful even in manufacturing and warehouse environments. With RFusion, a robotic arm with a camera and radio frequency(RF) antenna attached to its gripper, one can easily fetch their lost items.

The prototype developed by researchers relies entirely on RFID tags. Radio Frequency Identification tags, abbreviated as RFID tags, use electromagnetic fields to automatically identify and track tags attached to objects with the advantage of being cheap and battery-less.The RF signals can travel through most surfaces, and hence the robotic arm can identify items within a pile.

The robot uses machine learning to find out the precise location of the object, removes all the items enveloping it, gets a hold onto the object and further verifies if the correct object has been picked up. RFusion works in practically any environment as the camera, antenna, robotic arm are all integrated.RFusion can be extended to more critical applications like identifying and installing an auto-manufacturing part or helping an elderly individual perform daily tasks.

PRINCIPLE WORKING

To identify a spherical area around the tag attached to an object, RFusion uses its antennae which bounces off signals from the tag. It then combines the sphere with the camera input, which zeroes in on the object’s location. One problem associated with this is the fact that it is slow and inefficient. Although the robot gets a general idea of where the object is, it would have to swing its arm in all directions and take additional measurements to get its precise location. 

To optimize the robot’s trajectory, the researchers implemented reinforced learning to train a neural network. The agents are allowed to make mistakes or do something right and then rewarded or punished accordingly. If the number of moves in finding the location is reduced, RFusion is rewarded. 

The combined use of RF and optical information provides information on how the robotic arm should hold the object in terms of the angle of the hand and the gripper’s width. However, RF measurements and visual information cannot be relied on individually as it might result in an outlier. 

According to findings, when RF measurements are summarised, and visual data is limited to the area right in front of the robot, RFusion achieves a 96% efficiency in locating an object completely hidden under a pile.Beyond industrial applications, RFusion can be implemented in future smart homes.

RFusion will be a significant breakthrough for robotics operating in complex supply chains where identifying and picking the correct item is necessary. The researchers hope to increase the system’s speed and allow the robotic arm’s smooth movement in the future.

Source: https://news.mit.edu/2021/robot-finds-items-camera-antenna-1005

https://www.youtube.com/watch?v=iqehzw_aLc0