Thursday, December 6, 2007

Seeing in the Blind Spot

Every driver knows—or should know—about the “blind spot,” that part of the road that you just can’t see in any of your car’s rear-view mirrors, the spot that all too often hides a semi-tractor trailer full of hazardous wastes passing you at 70 miles an hour on a crowded turnpike.

R. Andrew Hicks, an associate professor of mathematics at Drexel University, has a better solution than a quick turn of the head and a yelp of surprise. He has created a slightly and elegantly curved mirror that provides drivers with a 45 degree field of view on the driver’s side. A flat mirror provides less than 20 degrees. The difference is dramatic.

Flat mirrors do not provide a wide enough field of view. Trucks and buses make use of spherical mirrors, which broaden the field, but increase the distortion of the image. The passenger-side mirror on a car, the one with the worrisome note that “objects in the mirror may be closer than they appear,” also trades a wider field for a distorted view.

Hicks has been working on creating very wide fields of view, in the new field of “omnidirectional vision,” sometimes also called panoramic vision, which has been developed to help robots to “see.” A mobile robot might use a camera with an extremely wide-angle “fish-eye” lens or with curved mirrors mounted in front of a standard lens. Omnidirectional vision provides a very large field of view.

Curved mirrors can provide the same kind of panoramic views—such as often seen in CCTV security cameras. Hicks was designing such mirrors for robots as a postdoctoral fellow at the General Robotics, Automation, Sensing, and Perception Laboratory at the University of Pennsylvania, where he received a Ph.D. in mathematics in 1995. He also was learning a lot about the problems of building them. The theoretical design of a curved mirror that has a wide and undistorted view is a “problem of classical optics,” Hicks says. That elegantly drawn curve, however, was until recently impossible to translate into a physical mirror. “The tricky part is that the machines that could actually build a practical design didn’t exist before 2000. So nobody really explored making such a mirror.”

Hicks can now make and demonstrate such a mirror, but he can’t sell it: U.S. law prohibits a curved mirror on the driver’s side. (It is allowed in Europe and Japan and as an add-on in the U.S.) Until that can be changed, he still advocates, as your driver-ed teacher did, an occasional quick glance over the shoulder.

###

Wednesday, December 5, 2007

18th-Century Architecture: 21st-Century Technology

All the students in the Digital Media Program at the Antoinette Westphal College of Media Arts and Design learn how to create the latest in computer graphics and animations. But the students working at Independence National Historical Park with Glen Muschio, the director of the program, are also getting a crash course in American history. Philadelphia may be America's most historic city, but much of what the tourist sees has been reconstructed over the centuries. Drexel University researchers are beginning to uncover what 18th-century Philadelphia was really like—and to create an interactive computer model of it that anyone can explore. Muschio, Chris Redmann, assistant professor of digital media and their students have been recreating exact 3-D replicas of colonial-era structures. The project brings together faculty and students from the computer science department in the College of Engineering, the culture and communication department in the College of Arts and Sciences, the College of Information Science and Technology and the School of Education.

The collaborative 3-D Colonial Philadelphia
will serve as a research and production center for models, animations and interactive media—a repository for "virtual artifacts" of colonial life, in searchable databases for researchers and scholars. The Dexter House was recreated based on information regarding the house foundation obtained at the site and a survey of buildings from the time period fitting the style of the house and buildings created by the Dexter House's carpenter.

"We want this to be a resource for research, production, use and evaluation of digital assets for studying U.S. history," says Muschio. The project is part Second Life, part social history, part SimCity, part computer science and part teaching tool. "What we hope to do is create a 3-D environment for teaching and learning about colonial American history in schools, at historic sites and on the Web" he adds.

One of the first homes recreated in their virtual 18th-century Philadelphia is a two-story brick house located in the vicinity of Fifth and Arch Streets that belonged to James Oronoko Dexter, a prominent and free African-American. Dexter is believed to have held meetings at his house for the creation of the first independent African American church. The original house was demolished in the 19th century, but recent archeological digs brought to light information about the house and site.

Muschio and his students have built the exterior and interior spaces of the home with historical accuracy and will scan artifacts found from the dig to restore the interior to its 18th- century look. Another building, the Whitall House, home of wealthy Quakers, which still stands in Woodbury, N.J. opposite Philadelphia International Airport, has been another model.

The future of the project includes regional sites and events that were historically significant to the development of the city's culture and economics. "3-D Colonial Philadelphia will evolve as technology advances and as we too develop cutting- edge technology to move the field forward," says Muschio. He wants to recreate the 18th-century city and its economic ecosystem, eventually populating its buildings with digitized characters, or avatars, who will show visitors the objects found there, and tell them about the significance of the site—like virtual tour guides. "As AI [artificial intelligence] capabilities develop, we hope that the interaction will get to the state where we can really do this on a big scale and people can get involved in role-playing games. That's where we're headed. Right now, were building the spaces in a photo-realistic way."

"We are working with the computer science department, which is interested in creating algorithms that will allow them to work with the National Park Service to virtually reconstruct the Dexter artifacts, based on the remnants that were uncovered," said Muschio. "We're also working with information-science technology to create databases so we can get a handle on all of this stuff, because we hope to make this available to other researchers in other cities who might not have the types of resources we have in terms of window types, brick face, things like that."

Chris Redmann, an architect and 3-D animator with an interest in historic preservation, is directing the construction of the 3-D virtual environments. He's teaching his undergraduate and graduate students to use the Historic American Building Survey (HABS) and the Historic American Engineering Record (HAER) as guides for reconstructing 3-D models of historic buildings. These digital assets will be used to construct virtual period buildings that are known to have stood on specific plots, but are undocumented in terms of detailed descriptions. The faculty and students will also research insurance, tax and deed records noting physical descriptions and positioning of structures.

The resulting database of colonial Philadelphia doors, windows, brick faces and other common features could be dropped into a scene or a row of houses quickly. Whole blocks, which might have sketchy historical records at best, could then be recreated with some measure of authenticity.

Working from HABS blueprints, undergraduate student Brian Gadomski, has already built an archive of 18th-century house doors and windows. "I'm bringing them into 3-D and I'm tracing them in the computer, and building, based on the original drawing, a geometry that matches exactly with those," he says. "It's a process that's relatively simple but will allow us to create a data base of doors and windows that can be used eventually in the complete Colonial Philadelphia."

"Using this method, it will not be possible to know what every specific house in the city looked like," said Muschio. "Detailed treatments will be reserved for historically significant structures that are well documented."

"I actually learned a lot about the Dexter house while doing it," said Chester Cunanan, a graduate student in digital media. "How they had the first meetings there that eventually led to the first African American church in Philadelphia and how the Quakers worked with that."Cunanan says that he has learned more than he expected. "You have to learn about the history of the place to build the place properly, and then the history catches you at that point. So when it starts, maybe you weren't totally into the history, but when it ends, you're not only interested in the place, but the history and the place. The history pushes you on—it's the impetus that drives you after the fun of playing with a new toy wears off."

###