It is very difficult, on a crowded multi-lane highway, to be sure that all road use is detected and that drivers don't escape payment. Hertzberger says several situations frequently cause problems for detection systems. Certain vehicles or combinations of vehicles may be misinterpreted by the system or overlooked entirely. For instance, he says, "Are you able to detect all motorcyles? Is someone able to avoid paying by driving behind a big truck? Can the system distinguish a car with a van behind it from a truck?"

When entertaining bids for the system, the Dutch government wanted a technology that would help it evaluate bidders' proposed software models of their systems. In response, UvA built a simulation harness, called Automatic Debiting System SIMulator (ADSSIM), that allowed contenders to implement models of their solutions. MatLab, the widely used mathematical problem-solving software from The MathWorks, Natick, Mass., was used to analyze the results. These tools were used to simulate the performance of the systems under different problem situations. At the next stage of the bidding, several companies were invited to build prototypes, which included sensors and video cameras. To test the prototypes, information recorded by the sensors could, for instance, be checked against the information recorded by cameras.

This complex data was stored in Matisse for analysis and examined using MatLab, via a Matisse-to-MatLab interface built by the University. As a result, the University discovered several situations where drivers would be able to avoid payment; the feedback was used by the bidders to improve their systems. "We could analyze data faster and better than anyone else, because we had the right tools," Hertzberger says.

Hertzberger notes that the combination of Matisse's complex data storage capabilities with MatLab, a package used to solve a variety of problems in research and engineering and other fields, could prove useful for people in other industries. "You have the possibility of storing experimental data and having a problem-solving environment for analyzing the data," he says. "For instance, we used this combination to do paint analysis of old Dutch paintings such as Rembrandt's "The anatomy lesson of Dr.Tulp."

Read about the Virtual Laboratory