Xerox's Big Data Analytics Catches Traffic Scofflaws

The next time you pull up behind a school bus that's stopped by the side of the road -- its red lights flashing and stop sign extended--be sure to smile and wave. No, not at the kids on board, but at the bus's high-resolution camera, the one that automatically captures video of bad motorists who pass the bus instead of waiting patiently.

Xerox, best known for its printers, photo copiers and other digital imaging products, is expanding beyond its traditional document management milieu to develop new services for transportation markets. These applications include imaging systems that read license plate numbers at toll booths, adjust parking meter rates on demand and detect the number of passengers in a vehicle--handy for catching single-occupant scofflaws cruising in HOV (high occupancy vehicle) lanes.

These services, some of which analyze video of traffic and parking patterns, generate vast amounts of data and require innovate uses of analytics and on-site processing.

Xerox is applying its decades of expertise in imaging technologies to transportation systems that can benefit from real-time analysis of data. These new applications pose challenges to scientists and engineers, however.

Xerox research scientist Abu Islam said in a phone interview with InformationWeek that video is often too bandwidth-intensive to transmit in raw form from a remote transportation application back to a central server.

"Imagine that one camera has 4, 6, or 8 megapixels, and it's taking 30, 40, or 50 frames per second. And all of these input systems are out in the field. Sending this data back to a central system is very hard because of the bandwidth issue," said Islam, who leads a team of scientists and engineers, many of whom work on software and control systems for Xerox's various transportation projects.

The solution? Embedded local processing.

"Every camera system is looking for certain information. If we can process that information locally, then a lot of times we get away from the bandwidth issue," Islam said. "It's intelligent sensing. It has local intelligence. It only sends back information that is important."

Xerox's CrossSafe system is an example of how the company is applying its image-processing skills to new transportation products. A turnkey system with a strategically angled high-resolution camera installed on a school bus, CrossSafe records video of any car that passes the stopped bus when the safety stop sign is extended. It also captures the car's license plate, thereby making it easier for law enforcement officials to ticket the car's driver.

This automated solution is far easier than previous, time-intensive methods, particularly those that required a human to spend hours watching video shot by a school bus camera.

"That's crazy," said Islam. "We can process each frame and find out when the stop sign came out, and if there's a vehicle next to it. We automatically detect that."

Some of Xerox's transportation technologies, such as its new Merge parking management system that uses advanced data analytics to change (e.g., raise) parking meter rates based on driver demand, probably won't be welcome by the majority of motorists. And one of Xerox's prototype systems, which uses automated image processing techniques to identify the number of occupants in a moving vehicle, does raises the Orwellian specter of a Big Brother bureaucracy run amok.

On the plus side, Xerox's imaging technologies can help cash-strapped state and local governments by allowing them to reducing the number of police officers need to monitor HOV lanes, the company claims, and by allowing municipalities to raise metered parking rates for special events.