Solo commuters frustrated by snarled traffic have taken extreme measures to sneak into high-occupancy carpool lanes: costumed mannequins in passenger seats, dolls swaddled like babies–even dogs in bonnets. But a company called Vehicle Occupancy, based at Loughborough University, in Leicestershire, England, says that it has developed an infrared camera-mounted scanning system that foils 95 percent of such trickery.
Automated systems for counting passengers have a heightened appeal as many urban areas with crowded roadways contemplate occupancy-based tolls as a way to pay for new highway construction. “We are far better at looking inside the car than the human,” says Tim Ballantyne, Vehicle Occupancy’s director of business development. “We can see 95 percent of the people in the car, whereas a trooper sees just 65 percent.” Vehicle Occupancy’s system, dubbed Dtect, was tested against humans from a distance of 150 feet, with cars moving at about 50 miles per hour, Ballantyne says.
According to Ballantyne, Dtect uses a proprietary infrared technology that can identify human skin by measuring its water content and detecting hemoglobin in the blood just beneath the skin. Mounted at the mouth of the travel lane, the Dtect camera captures two images of oncoming vehicles at different frequencies in the infrared range. At those frequencies, live human skin has different light-absorbing attributes than heated plastic or steam emanating from a cup of coffee–or even a bladder of hot water, which might be used to try to fool the system.
In addition to recognizing the infrared signatures of hemoglobin and water, Dtect’s software can differentiate between a person and a large animal, using algorithms similar to those found in face-recognition systems. “There are two levels,” says Ballantyne. “First, we filter out things that are not live–the dummies and so on–and in the second layer, we look at the image and can reject the second level of trickery that might be dogs or a warm human hand held up to where the infrared light is aimed at.” Ballantyne says that skin tone and heavy makeup are not obstacles to the system, but a small child bundled in a back seat could be missed.
The system’s price–100,000 English pounds for two infrared illuminators, two cameras, a central processing unit, waterproof housing, software, and a battery pack–could make it a tough sell. But it has still attracted interest in Virginia.
Officials there are moving toward a multitiered pricing system for the tolls that will fund 14 new miles of high-occupancy-vehicle interstate roadway for the Capitol Beltway, and another 56 miles between Fredericksburg and the Pentagon. Solo drivers who opt to pay the toll will have access to the roads; carpoolers will ride free. But the state needs a way to distinguish cars with one passenger from those with multiple passengers without forcing vehicles to slow down to be observed by state police.
“You can’t just eyeball it to determine who is going in there illegally or legally,” says Jeff Caldwell, spokesman for the Virginia Department of Transportation. “You need an automated system that can operate day and night, independently, in all kinds of weather.”
Caldwell says that his agency is working with Fluor-Transurban, a joint venture between two multinational highway-building firms, to create the toll roads. Jennifer Aument, a spokeswoman for Fluor-Transurban in New York, confirms that her company is interested in the Dtect system.
“We are at the beginning point, but we see promise in that,” she says. “It needs intense and rigorous review. The hope is that this technology would be able to distinguish between a human, say, and a dog.”
Still, Caldwell says, “We have not capitulated to that as the appropriate technology.” Early word that Virginia was even discussing an infrared system for counting passengers, he says, prompted immediate privacy concerns among residents and their state representatives.
“Our people want to know how this fits with privacy worries, and how enforceable it will be,” Caldwell says. “This is a really cutting-edge project, and there are a lot of uncertainties.”
Ballantyne is aware of the privacy fears and says that Dtect can be programmed to scramble or obscure people’s faces. In prototype testing, he says, the system was able to superimpose a green circle over passengers’ faces. And the images themselves, he notes, are gray scale and not useful for identification purposes. “When we export the image of the number of persons in the car, we make it so it is impossible to identify the faces,” he says. “And of course if someone was hiding in the boot, we wouldn’t be able to detect them.”
As for good old-fashioned dummies in the car, “it gets the headlines, but it is very rare,” Ballantyne says. “People really do have a lot of better things to do with their lives.”
The big new idea for making self-driving cars that can go anywhere
The mainstream approach to driverless cars is slow and difficult. These startups think going all-in on AI will get there faster.
Inside Charm Industrial’s big bet on corn stalks for carbon removal
The startup used plant matter and bio-oil to sequester thousands of tons of carbon. The question now is how reliable, scalable, and economical this approach will prove.
The hype around DeepMind’s new AI model misses what’s actually cool about it
Some worry that the chatter about these tools is doing the whole field a disservice.
The dark secret behind those cute AI-generated animal images
Google Brain has revealed its own image-making AI, called Imagen. But don't expect to see anything that isn't wholesome.
Get the latest updates from
MIT Technology Review
Discover special offers, top stories, upcoming events, and more.