Researchers from MIT created a robot finger that can feel objects beneath a granular surface.

Robots are good at identifying objects in the open but are weak when it comes to subsurface detection. To find objects buried in sand or granular surfaces, a robot would need slender fingers that can wriggle out of the sand and are sensitive enough to feel the shape of the object.

In the past, researchers have used technology that senses the subterranean from the surface. These technologies only provide a hazy view of submerged objects and can struggle to differentiate objects, like rock versus bone.MIT researchers developed a "Digger Finger" robot that digs through granular material, like sand and gravel, and senses the shapes of buried objects. Source: Radhen Patel, Edward Adelson, et. al.MIT researchers developed a "Digger Finger" robot that digs through granular material, like sand and gravel, and senses the shapes of buried objects. Source: Radhen Patel, Edward Adelson, et. al.

The team created a sharp-tipped robot finger, named DiggerFinger, equipped with active sensing to meet the challenge of identifying buried objects. DiggerFinger can dig through granular material and accurately sense shapes. One day this robot finger could perform various subterranean duties like finding buried cables or disarming buried bombs.

The first challenge was to create a finger that is slender and sharp-tipped. While developing the robotic finger, the researchers used a tactile sensor called GelSight composed of a clear gel covered with a reflective membrane that deforms when objects are pressed against it. Behind the membrane are three colors of LED lights and a camera. The lights shone through the gel to the membrane and the camera collected the membrane’s pattern reflection. Computer vision algorithms extracted the 3D shape of the contact area where the finger touched the object. This device had a great sense of touch, but it’s bulky so the team needed to make adjustments.

To integrate GelSight into DiggerFinger, the team slimmed down the GelSight sensor in two ways. First, they changed the shape to a slender cylinder with a beveled tip. They also ditched two-thirds of the LED lights with a combination of blue LEDs and color fluorescent paints. These improvements compacted the device. The final device had a tactile sensing membrane that measures about 2 cm2.

Next, the team worked on motion by mounting the finger on a robotic arm and observing it dig through fine-grained sand and course-grained rice. As this granular media can jam and make it difficult for the finger to penetrate, vibration was added to DiggerFinger. Vibration tests at different operating voltages showed that rapid vibrations helped fluidize the media and clear jams. The device had an easier time unjamming rice than the sand.

The team also tested twisting motions in rice and sand. Sometimes the media would get stuck between DiggerFinger tactile membrane and the buried object. In rice, the trapped grains were large enough to completely obscure the shape of an object, but the jam could be cleared with a little wiggling. The trapped sand was harder to clear, but the small grain size allowed the finger to sense the general contours of the object. Operators will have to adjust DiggerFinger’s motion pattern for different settings depending on the given media.

The team plans to explore new motions so the robotic finger can navigate a variety of media. Read the team's paper on DiggerFinger here.