From surgery to robotics, touch is the key

By Beth Azar
Monitor staff

Surgeons performing microscopic surgery rely on visual cues from a video screen to determine the best place to make an incision, place a sponge or insert a suture. They have none of the tactile cues?the feel of a wiry vein, the edge of a muscle or the softness of tissue?that they have performing traditional surgery.

In large part, that?s because the design of instruments that give surgeons tactile input is in its infancy. Engineers and computer scientists are working on the problem, but truly effective tactile simulators won?t be available until researchers can answer some basic questions about how people perceive touch, says psychologist Susan J. Lederman, PhD, of Queens University in Ontario, Canada.

She and her colleagues are trying to provide those answers. She oversees Queens University?s Touch Laboratory and has collaborated for more than a decade on studies of touch with Carnegie Mellon University psychologist Roberta Klatzky, PhD.

Together, Lederman and Klatzky have amassed data on how people learn about objects through touch and what kind of information they obtain. They?ve also begun to team up with engineers to determine whether touch input is important for jobs such as operating robots from afar and performing microscopic surgery.

'If you don?t understand the capabilities and limitations of humans, you can?t design systems that permit them to operate effectively on remote environments?whether they be real or virtual,' says Lederman.

Research on hands

Research on the haptic sensory system focuses mostly on the hands because the fingertips contain one of the highest densities of tactile receptors, says Klatzky.

'The same kinds of explorations could be formed with other limbs or with the mouth,' she says. 'But the hands provide the dexterity and sensitivity we need for our experiments.'

Some of their earliest work set out to determine how people learn about objects with their hands. They found that people use six basic 'exploratory procedures':

? Lateral motion?rubbing the fingers across a surface provides information about an object?s texture.

? Pressure?pressing down on an object provides information about its hardness.

? Static contact?holding the fingers in one spot, provides information about an object?s temperature.

? Unsupported holding?holding an object out away from a support provides information about its weight.

? Enclosure?wrapping the hand around an object provides information about its global shape and volume.

? Contour following?moving the fingers about the perimeter of an object provides information about an object?s exact shape.

People perform these procedures in a logical pattern: First they grasp the object?a quick, crude way of gaining a lot of initial information. Then, if necessary, they begin to use the specialized hand movements.

These findings provide a basis for understanding how people intelligently explore objects, says Lederman. And if engineers want to design robots that can use tactile sensors to analyze an environment, they will need to devise systematic manual testing procedures that, when sequenced appropriately, will extract tactile information effectively. A robotic hand may not use the same exploratory procedures a human hand uses?its sensors may be different?but engineers can use Lederman and Klatzky?s findings to develop their own version of procedures the robots can use.

Materials, not geometry

Psychologists? research is particularly key to working on systems that enable humans to work in remote or inaccessible environments, such as microsurgery, radioactive sites and even other planets. Engineers must understand which tactile properties people are most attuned to.

Over the years, research findings converge on the principle that touch is extremely sensitive to material properties?how hard, cool, pliant and rough an object is. But the sense is relatively poor at determining spatial and geometric properties, such as whether an object is sloped to the right or left, or whether an edge is horizontal or vertical.

Indeed, in a recent series of studies published in the Journal of Experimental Psychology (Vol. 23, No. 6, p. 1680?1707), Lederman and Klatzky found that people process geometric and spatial properties much more slowly and less accurately than material properties or the presence or absence of edges. Furthermore, the brain can search for a material property with six fingers simultaneously but must search for a spatial target one finger at a time, the researchers found.

These findings suggest that engineers should design tactile interfaces that provide information about materials and edges rather than geometry, whenever possible.

'Tactile interface designers shouldn?t expect people to read fine patterns with their hands,' says Lederman.

However, people will quickly and easily respond to tactile cues such as surface texture or sharp edges. Indeed, adding texture or edge-orientation cues, rather than labels, to dials on a car radio could reduce the need for people to look down to change the station, says Klatzky.

The importance of force

Because the haptic system is so complex, adding tactile information to remote-access devices is a daunting prospect for engineers. So it?s up to psychologists to convince them there?s value in adding tactile feedback to instruments, says Lederman. She and Klatzky have preliminary evidence hinting that feedback does provide a certain advantage.

They tested people?s ability to perform several tasks with and without feedback to their index fingers. The tasks were chosen to represent situations that people might encounter when using a remotely operated instrument. For example, they measured people?s ability to feel vibrations; to sense whether they can feel two distinct objects or just one; and to detect the presence of a thin nylon hair. They also tested perceptual abilities, including the ability to judge how rough a surface is and to compare the roughness of two surfaces. And, they assessed people?s ability to detect a rigid mass embedded in simulated tissue.

To simulate a no-feedback situation, the researchers covered participants? fingertips with a rigid fiberglass sheath. The sheath had a dramatic impact on several aspects of sensitivity and perception, Klatzky and Lederman found. People?s ability to sense the thin hair declined by 73 percent and their ability to detect two objects as opposed to one declined by 321 percent. They also lost the ability to tell whether two objects were oriented in the same direction and had far more trouble detecting a mass embedded in tissue.

'The results suggest there may be significant perceptual costs when spatially distributed fingertip forces are not sensed or displayed to novice operators of teleoperator and virtual environment systems,' write Lederman and Klatzky. Their article is in press in Presence, the Massachusetts Institute of Technology?s magazine on research related to virtual reality.

Maximizing perception

The key to designing more useful virtual environments and teleoperator devices is to supply people with a blend of sensory information, says Lederman. In the real world, people tend to integrate information from all their senses rather than operate with one sense at a time.

'For example, touch and vision complement each other,' says Lederman. 'They do different things very well.'

While vision provides information about an object?s geometric features, touch is unparalleled in its ability to extract information about materials. For a surgeon trying to decide where to begin excising a patch of cancerous tissue, it might be helpful to feel the texture and compliance, and not just rely on the shape.