MIT’s Robotic Finger Identifies Buried Items

Robots are great, it seems like they can do anything. But one thing they still struggle with is identifying buried items. Granular materials like sand make it difficult to feel the object and can cause jamming. MIT researchers recently developed a new robot to overcome this challenge. The team’s new robot, Digger Finger, is a slender robotic finger that can dig through granular material and sense the shape of the buried objects.

The robot is based on the researchers’ previous tactile sensor called GelSight. The sensor includes a clear gel covered with a reflective membrane that deforms when objects press against it. The membrane consists of three colors of LED lights and cameras. The lights shine on the membrane while the camera collects the membrane’s pattern of reflection. Computer vision algorithms detect the 3D shape of the area where the finger touched the object.

Digger Finger features an upgraded version of GelSight, which is slimmer with a slender cylinder shape and a beveled tip. Two-thirds of LED lights were removed in favor of a combination of blue LEDs and colored fluorescent paint. It also has a tactile sensing membrane of about two square centimeters or about as big as a fingertip.

For testing, the device was mounted on a robotic arm, which allowed it to move and dig through fine-grained sand and coarse-grained rice. This type of material normally jams, making it difficult to dig through. But Digger Finger was able to twist and turn the grains with its built-in vibrating motor. Though trapped sand was still difficult to clear, the robot could still feel the general contours of the object. The team hopes to optimize Digger Finger’s mobility by exploring new motions.

Though there’s still a lot of tweaking that needs to be done, the team already has a number of practical uses in mind. Digger Finger could be used to find land mines, check data cables for cuts, or even find buried bombs without endangering people.