Robotics

Robots with "eyes" in their hands may prove more dextrous than others

Robots with "eyes" in their hands may prove more dextrous than others
Researchers at Carnegie Mellon University have developed a three-fingered robot hand that's extremely sensitive to touch and pressure thanks to built-in optical sensors
Researchers at Carnegie Mellon University have developed a three-fingered robot hand that's extremely sensitive to touch and pressure thanks to built-in optical sensors
View 4 Images
Researchers at Carnegie Mellon University have developed a three-fingered robot hand that's extremely sensitive to touch and pressure thanks to built-in optical sensors
1/4
Researchers at Carnegie Mellon University have developed a three-fingered robot hand that's extremely sensitive to touch and pressure thanks to built-in optical sensors
One of the new flexible sensors with its LED light source
2/4
One of the new flexible sensors with its LED light source
Wave patterns show the reaction of a flexible sensor to movement
3/4
Wave patterns show the reaction of a flexible sensor to movement
The three-fingered optic sensor equipped robotic hand close up
4/4
The three-fingered optic sensor equipped robotic hand close up
View gallery - 4 images

en we think of robots, we all too often anthropomorphize them by giving them eyes in their heads, fingers on their hands, and toes on their feet. But just because this is the way humans evolved doesn’t make it ideal. Robots with eyes where they need them most, for example, could be much more efficient than just having them restricted to one place. In this vein, researchers at Carnegie Mellon University (CMU) recently developed a tri-fingered robotic hand with numerous inbuilt optical detectors to act as adjunct sensors. At the same time, they also fashioned a new type of stretchable optical sensor to accompany such devices.

CMU researchers used optic fibers to connect 14 strain sensors in each one of the three fingers of a soft robotic hand, enabling it to determine the location of the fingertip’s contact and the force being applied, down to a measure of less than one-tenth of a newton.

"If you want robots to work autonomously and to react safely to unexpected forces in everyday environments, you need robotic hands that have more sensors than is typical today," said assistant professor of robotics at CMU, Yong-Lae Park. "Human skin contains thousands of tactile sensory units only in the fingertip and a spider has hundreds of mechanoreceptors on each leg, but even a state-of-the-art humanoid such as NASA’s Robonaut has only 42 sensors in its hand and wrist."

Mimicking the bone structure of human appendages, the fingers on the CMU robot hand have a fingertip, a middle nodule and a base node all joined with flexible joints. Each of these elements has been created with a 3D printer in rigid plastic and incorporates eight fiber Bragg grating (FBG) detectors, which sense force by measuring alterations in light wavelengths reflected by the optical fiber.

In turn, the three finger sections are sheathed in a soft silicone rubber "skin" which has a total of six sensors implanted in it that detect the point where contact with a grasped object has been made. To bend each of the fingers, an elastic "tendon" is pulled in, while straightening each finger relies on a passive elastic tendon applying opposing force when the bending force is released.

The three-fingered optic sensor equipped robotic hand close up
The three-fingered optic sensor equipped robotic hand close up

Carnegie Mellon researchers are well-versed in the creation of soft robotics, with malleable bodies, composed of fabrics, inflatable sections, and light plastics that use pneumatics for movement rather than hard gears and rigid pulleys. Their creations are a much more user-friendly attempt at introducing robots as caregivers and home helpers rather than heavy, often dangerous industrial machines.

However, because their robots are much more flexible than conventional automatons, the sensors that they incorporate in them must also be more giving, bendable, and stretchy. As such, adding standard pressure or force sensors to flexible robots can add their own problems because conventional conductors are susceptible to breaking, and wires are prone to interference from other electromagnetic sources.

Conversely, an optical fiber may house several sensors on its own. Because of this, all 14 of the sensors found in the fingers of the CMU robotic hand are connected with just four optic fibers that are not only unaffected by electromagnetic interference, but are much more flexible. Despite this innate flexibility, assistant professor Park says that optical glass fibers do not stretch, and polymer ones barely cope with much more than 20 to 25 percent elongation before they fail. This means that one entire area of a hand's range of movements is limited. To overcome this, Park previously tried microfluidic membrane force sensors, but these tended to leak.

As a result, Park collaborated with CMU mechanical engineering student Celeste To and University of Texas mechanical engineering student Tess Lee Hellebrekers to create a very stretchable, flexible optical sensor, formed from a combination of commercially available silicone rubbers. This sort of flexible optical sensor could also be embedded in soft skins. Skin created in this way would be able to detect contact and force, like the optic sensors in the CMU hand, but would also be able to stretch longitudinally in a way that optical fibers cannot.

The Carnegie Mellon researchers, along with researchers at Intelligent Fiber Optic Systems Corporation, who helped develop the hand with support from NASA, recently discussed their work at the 2015 IEEE International Conference on Intelligent Robots and Systems in Hamburg, Germany.

The short video below shows the three-fingered robot hand in action.

Source: Carnegie Mellon University

Multi-Fingered Robotic Hand

View gallery - 4 images
1 comment
1 comment
jochair
You could consider a robot with three arms, the third with eyes, that can be brought close to objects