[Architectural Record 2000]
Imagine a building that can sense when an earthquake strikes and can instantly transform itself into a flexible, treelike structure capable of dissipating the seismic shock. Or imagine a bridge that can sense when it's about to be ripped apart by hurricane-force winds and can instantly stiffen its support system to withstand the assault. Or imagine a car that can detect when its engine has developed a potentially fatal flaw and can reflexively adjust its operations to mitigate the danger. Feng Zhao, a computer scientist at Xerox's Palo Alto Research Center (PARC), has been imaging such ``smart structures'' for a long time, and he says the day may not be far off when they are part of our everyday lives.
Zhao and his colleagues in the Collaborative Sensing Project at Xerox PARC are developing a system composed of tiny, inexpensive sensors that can capture information about their immediate environment--such as heat, light, and motion--and share that data with thousands, or even millions, of identical sensors scattered nearby. That network of ``massively distributed sensors'' can analyze the resulting mountain of information in real time and initiate an appropriate response almost instantaneously. In the case of an earthquake-resistant building, that response could take the form of sending electrical signals to a smart material called a ``shape-memory alloy'' embedded throughout in the building's structural frame. That material, in turn, would respond to the signals by expanding, thus providing controlled flexibility to the structure and enabling it literally to wiggle through a seismic event. Conversely, ``collaborating sensors'' in a wind-resistant bridge or building could send electrical signals to a smart material in the span's support structure, prompting the material to contract, thereby stiffening the bridge's resistance to wind.
“Think of these sensors as intermediaries between the physical world—the world of buildings and bridges—and the human world, the world of knowledge and ideas.”
``Think of these sensors as intermediaries between the physical world--the world of buildings and bridges--and the human world, the world of knowledge and ideas,'' Zhao says. ``In one direction, they can take information from the physical world and turn it into something that humans (or computers) can look at and make sense of. In the other direction, they can infer human needs by capturing contextual information about people's preferences, and tell the physical world how to respond.''
What distinguishes collaborating sensors from their dumber cousins--such as thermostats or barometers--is that they contain sophisticated software that enables them to organize and share information in an almost organic way. ``They're like the neurons in our brain,'' Zhao says. ``They respond adaptively to the environment. There is no central database. Each node [sensor] has a little bit of memory. Groups of nodes can form supernodes that have more knowledge about the environment.'' And, like the human brain, the sensors are very good at filtering out unimportant information. ``Most of the time they're in sleep mode. They only wake up when something significant happens,'' he adds.
Collaborative sensing is already being used in a limited way in some everyday devices. Xerox printers, for example, contain sensors that can detect tiny deviations in the movement of paper through the machine and can automatically make adjustments to avoid jams. But Zhao says the sensors will have to get a lot smaller, smarter, and cheaper before they can be deployed on a massive scale in construction and other commercial applications.
The Xerox team has developed fingertip-sized sensors that could be mass-produced at a cost of about $5 or $10 each. Within five to 10 years, the company expects to shrink the sensors to the size of peppercorns or smaller while, at the same time, dramatically increasing their capacity to process the blizzard of raw data generated by a massively distributed network (a mountain of miscellany, Zhao says, that will make the amount of information available on the Internet seem small in comparison). But it may take another 10 years beyond that before the cost of the sensors will be low enough--pennies apiece--to make economic sense for builders. At that point, Zhao says, collaborating sensors will be not much different from bulk-building materials such as nails and bolts: They could be embedded in steel beams and other structural elements or stirred into paint or asphalt, and sprayed onto walls or roads.
