Friday, 22 June 2012 11:16
June 22, 2012
Understanding how humans put together their picture of the world has always been complicated simply because there are so many different sources of input that people rely upon. Just as importantly, humans all have finely developed strategies to make the best use of the information they have available to them.
Researchers at the University of Southern California believe they have developed a method of applying some of these strategies to give robots access to some of the crucial information that many people often take for granted: a sense of touch.
While robots are used extensively for material handling, particularly in the manufacturing sector, traditionally most of these robots either require limited sensory inputs to function properly or use visual signals. A range of engineering tools and carefully designed environments minimize the need for this information, but adding in a reliable sense of touch could expand the range uses to which robotic equipment could be put.
The team from USC's Viterbi School of Engineering, led by professor of biomedical engineering Gerald Loeb, hopes they might have found a way to give robots another sense by adapting both structures and strategies developed over millions of years of evolution in humans.
The first step was to give the robots sensations similar to what a person might find in their hand. For this, the researchers created a system known as BioTac, which incorporates an electric sensor known as a hydrophone, capable detecting vibrations, into a solid core that is surrounded by a liquid encapsulated in an artificial skin. This skin features ridges like artificial fingerprints that vibrate when rubbed against materials, just as occurs in a human's fingers. These vibrations are then carried through the liquid filling to hydrophone.
Improving on nature
Creating a working finger, however, is not the same as giving robots a sense of touch. Even with all the same tools as a human, hands are useful at least in part because there are certain "exploratory movements" that humans use to get the most information out of the signals from their hands.
"Humans have been shown to be good at using active touch to perceive subtle differences in compliance," the researchers explained in Frontiers in Nuerorobotics, where the results of their research were published. "They tend to use highly stereotypical exploratory strategies such as applying normal force to a surface."
The team developed a means of using the new BioTac sensors by expanding upon the ideas of the 18th-century mathematician Thomas Bayes who first characterized these typical motions. Ultimately, they created a system that was able to distinguish between 117 different materials, just simple samples taken from local fabric, stationery and hardware stores. With some training, the robot managed to reach 95 percent accuracy, even higher than humans. What's more, similar materials that completely stumped humans only provided the occasional problem for USC's robotic hands.
Prosthetics and sensitive machines
The research behind the BioTac system started primarily with a grant from the Keck Futures Initiative of the National Academy of Sciences, intending to develop more advanced prosthetics. These systems would provide greater information and greater control to amputees.
However, the ability to create touch-sensitive robots could ultimately prove to have broad applications, since the system was found to be effective not only in identifying the texture of materials but in determining its hardness. Immediate applications could be limited to various testing devices, but this ability could also be useful in a variety of drones and potentially manufacturing equipment.
|< Prev||Next >|