The RoboMote: A platform for research in robotic sensor networks... Credit: USC Robotic Embedded Systems Lab (Gaurav S. Sukhatme, Director )

Many kinds of sensors are available to detect information about the physical world, ranging from thermometers for temperature to chemical sensors for pollutants to seismometers for earthquakes. Robotics research, at the Center for Neuromorphic Systems Engineering and elsewhere, is leading to the development of intelligent sensors that can adapt to changing environments or move to better locations to collect the best data.

Tiny, two-wheeled and wireless, RoboMotes are designed to create networks of sensors that move and reconfigure themselves to adapt to new situations. Developed at the University of Southern California's Robotic Embedded Systems Lab, each golf-ball-sized RoboMote includes a wireless network interface, two wheels with odometers; a solar cell for power; a compass for direction; and bump and infrared sensors for obstacle avoidance. Because of their small size and low cost, RoboMotes make it possible to experiment with larger numbers of sensors in dynamic networks.

A robotic sensor node from UCLA's Network Infomechanical Systems (NIMS) project... Credit: NIMS, UCLA

The Networked Infomechanical Systems (NIMS) project at UCLA is deploying robotic sensors suspended from lightweight cables. NIMS research is giving the sensors the intelligence to assess their own performance, reposition themselves along the cables and select the best sensing tools for each scientific task.

In the James San Jacinto Mountain Reserve near Idyllwild, California, a NIMS robotic sensor suspended between two trees has endured rain, snow and sun while its camera monitors plant growth throughout the experimental zone. Future test networks will observe a mountain stream ecosystem--from the ground to the treetops--for global change indicators, and monitor coastal wetlands and urban rivers for biological pathogens. The same technologies could one day be applied to securing public infrastructure.

Robotics researchers are also looking at developing novel sensor capabilities. For example, the sense of touch embedded in the human skin has been difficult to duplicate. Human skin bends and stretches to let us move around, but electrical circuits don't like to be stretched. Sigurd Wagner and Stephanie Lacour of Princeton University have created a robot "skin" with the first stretchable metal film interconnects for elastic integrated circuits.

Consisting of stripes of gold 25 nanometers thick on a silicone membrane, these interconnects can conduct electricity while being stretched at least 15 percent. Some samples have been stretched to twice their original length and remained conductive. But so far, the researchers don't know why this technology works.

Another problem for robots is that space is limited. All the sensors needed for many applications won't fit into the robot if they're any bigger than a speck of dust. NSF has funded many projects in microelectromechanical systems (MEMS), which have provided new sensors relevant to robotics. Micromachined artificial hair cells, for example, might provide the ability to sense wind blowing or liquid flowing from different directions (see "Robots and Biology").