Dr. Dobb's Journal May 2003
This year's Lemelson-MIT Student Prize for Inventiveness, which carries a $30,000 award, was bestowed on graduate student James McLurkin, for building more than 100 small robots and programming them to behave as a swarm. The robot "bees" are designed as cubes about 4-inches square, and equipped with wheels, bump sensors, light sensors, self-charger, radio modem, and audio system. They keep track of their positions relative to each other, and cooperate in such tasks as following a leader, clustering, or patrolling a perimeter. McLurkin plans to add food sensors, trail sensors, and cameras to each robot to make the swarm more capable.
McLurkin began his work as an MIT undergraduate, where he built 12-inch cubic robots designed to simulate the behavior of ants. (MIT calls these "the world's smallest self-contained autonomous robots.") He then obtained his master's degree from the University of California, Berkeley, and expects to receive his Ph.D. from MIT in 2006. He also works as lead scientist for iRobot Corp. (http://www.irobot.com/), makers of the Roomba robot vacuum cleaner. His robot swarm was assembled at iRobot.
The U.S. Department of Commerce's Technology Administration has released two reports (http://www.ta.doc.gov/space/ library/reports/#data-2002) on the commercialization of space: "Space Economic Data" and "Market Opportunities in Space: The Near-Term Roadmap."
According to the Commerce Department, launch costs remain one of the biggest hurdles to commercialization of space technologies, but the Department believes that space could still pay off, even in the short term. Industries that require a pristine environment, such as semiconductor manufacturing or delicate biotechnical research, are beginning to look at space-based operations. Their research is hindered, however, by the lack of solid data on the economics of space.
Estimates from different sources on the growth of the space sector vary widely. Different statistical agencies use different definitions of "space," and industries often do not separate out data on their space ventures from that of their more conventional projects. In fact, much of the current technology is hybrid, making it difficult to obtain specific data on commercial space ventures. The "Space Economic Data" report urges the Department of Commerce to promote reporting and standardization of space economic data by sending out surveys, with the goal of ultimately creating and maintaining a publicly accessible clearinghouse of information on the space sector.
Embedded Linux vendors have teamed to create a common API specification to ensure interoperability and application portability across their products. The Embedded Linux Consortium Platform Specification (ELCPS) now covers only core Linux services, but additional APIs will be defined in the future. The 124-member Embedded Linux Consortium (ELC; http://www .embedded-linux.org/) is also working on conformance test policies, which will become the basis for test suites.
The ELC is considering turning over the ELCPS to an international standards body. There is also a proposal within the ELC to change the ELCPS's name to "Embedded Standards Base," to reflect symmetry with the Free Standards Group's Linux Standards Base.
In a recommendation to the U.S. Congress, the Department of Commerce's National Institute of Standards and Technology (NIST) has embraced biometric technology as "mature," recommending that Congress adopt both fingerprint and facial recognition systems to identify visa holders at U.S. borders. "Facial recognition is a viable technology for verification or 'watch lists' but not for large-scale identification," the Institute concluded (see http://www.itl.nist.gov/iad/894.03/NISTAPP_Nov02.pdf). "Our measurements indicate that a dual biometric system including two fingerprint images and a face image may be needed to meet projected system requirements for verification."
NIST's evaluations were based on fingerprint testing performed on an Immigration and Naturalization Service database, consisting of 1.2 million prints from 620,000 individuals, and facial recognition tests conducted using a Department of State database of 121,000 images from 37,000 individuals. A 90% probability of true verification, with a 1% probability of false verification, could be achieved by fingerprint matching, as long as at least four finger impressions were available. The best facial recognition systems achieved the same level of accuracy, though under imperfect outdoor conditions, facial-recognition accuracy dropped to 47%.