At first glance, with its wave pools, wind tunnels and virtual-reality rooms, Canada’s National Research Council (NRC) has a lot in common with an upscale amusement park.
The difference is these devices are used in the name of research and learning, rather than in fun. According to one scientist at the Ottawa-based facility, however, research is fun – if it’s done right.
A visit to the Institute of Information Technology (IIT), one of the NRC’s many Ottawa-based groups, reveals work with Canadian companies in five major areas of research: visual information technology, software engineering, integrated reasoning, network computing and interactive information.
Visual Information Technology
The Visual Information Technology Group develops and applies 3D imaging and modeling techniques for visual communication and industrial automation applications.
According to Marc Rioux, project leader of 3D sensors for the Visual Information Technology Group, the group recently developed a colour laser scanner which can digitally record exact copies of objects pixel by pixel, by coupling an RGB laser to the optical fibre of a digital camera. The 3D sensor, laser scanner ‘camera’ employs a laser beam which is scanned on the surface of a given object, and the return signal gives the co-ordinates in space of that object, he explained.
“So the 3D files have been used for demonstration and documentation of objects, like in a museum context, and the future of this application has much potential. Because you can now record the shape and colour of objects,” Rioux said.
One of the main objectives of this process is to record changes to objects over time, he explained. Up until now, an artefact’s rate of change was virtually impossible to measure accurately.
“Now, the same object can be rescanned a few months later, or a few years later, to monitor changes. So if a piece in a museum is exposed to an environment which is causing damage that makes it change in colour, those modifications can be recorded with the technique we have developed.”
The capability for 3D digitizing started in 1981, and was used specifically for reverse engineering – making a CAD file from an existing physical object. 1990 saw the addition of colour capabilities to the digitizing process.
“We went to Israel and digitized a tomb,” Rioux said. “Now you can visit that tomb at the same scale using a large screen. You can enter it, look around and visit it as if you were on the site. And a lot of the sites around the world are now closing, because too many tourists have damaged them.”
The realism produced through digitization is far superior to photographs pasted on flat walls, he said.
“The human observer is very sensitive to the difference between synthetic and real. That’s why we like to call it ‘virtualized reality’ instead of virtual,” said Jacques Domey, group leader of the Visual Information Technology Group.
“When we digitize an object we have to take views all around in order to cover the complete surface of the object. So there is a need for software to be able to integrate all the multiple views and generate that single model of those multiple views, while eliminating the redundancy of points, and also compress the model so that it can be viewed on a very low-cost workstation,” he said.
“The concept is that we acquire a very detailed model that is used as a reference, and then out of this very high-accuracy model, we can generate lower levels of accuracy that will suit any application.”
The original is always retained on file, he said. “You can scale down the resolution so that anybody can visualize it on the Web, but you still have the high resolution records if you want to study deterioration and changes over time.”
The group has also created a search engine, called Nefertiti, which uses geometric shapes and 3D files rather than text, Rioux said.
“You put a shape in the window and it will search out the best match for the shape in the database. So it’s really the same concept (as standard engines) but all based on 3D objects.
“We foresee a very large potential application in electronic commerce,” he explained. “We are convinced that companies will start putting 3D models and scanned objects on the Web for consumers, and our visual interface and 3D search could be instantly used in other countries because it is not language-dependent.”
The search engine was named for the oldest colour sculpture known. It was believed that the artist made a copy to reduce the number of sittings needed. “So it was a 3D colour copy, which is exactly what we are trying to do 5,000 years later,” Rioux said.
The Software Engineering Group creates tools and techniques to help Canadian software companies enhance products and development processes, with research focusing on the entire lifecycle of products.
Recent areas of research include automated detection of code duplication in multiple code bases, identifying ways to reuse software components, configuration management, real-time and embedded systems and human/computer interaction. The goal of the research is to allow developers to increase productivity, shorten development cycles and reduce maintenance costs, while at the same time enhancing the quality of software, explained Anatol Kark, group leader of the software engineering group.
“Software engineering, at the moment, is not quite an engineering discipline. What we are trying to do is bring the practice of developing software to the engineering level,” Kark said. “And we do that by looking at methodologies and practices, or designing new ones.”
One main project involves research in building predictive models of software. “We want to be able to say how the software is going to behave under certain conditions. That is especially important in real-time systems,” he said.
“Part of the problem when you develop software is that you should be able to say how much time it will take to execute it, how much memory it requires and how it is going to interact with the environment. Very often, if not most of the time, people just don’t know.”
Kark said it always drives him crazy when a developer finishes writing a program, then runs it to see what will happen. “My statement was always ‘You should know what’s going to happen – because you wrote that piece of code.’ But I understand that it is not always possible – there are many dependencies on it that you have no control over.”
The group has recently developed an environment in which to study real-time behaviour of Java applications. Once these behaviours are mapped, the hope is to build a predictive model of the language using this information, he said.
In addition to research, the group is also involved in the writing and reviewing of areas of knowledge for various software standardization activities and bodies, such as ISO. This type of work, “because it codifies what a software engineering professional should know,” is crucial to the industry, Kark said.
Much of the Integrated Reasoning Group’s research is targeted at the design and development of automated diagnostic systems for use in many diverse areas, including commercial airlines, paper making, aerospace, data mining and network management.
One project, the Integrated Diagnostic System (IDS), is working to develop advanced software-based diagnostic techniques for the maintenance of mechanical equipment. Initially IDS is being used in collaboration with Air Canada and General Electric to categorize automatic, real-time, aircraft-generated messages and to diagnose mechanical problems, said Francios Dub