How New Indoor Navigation Systems Will Protect Emergency Responders

Tracking firefighters in blazing buildings helps keep them safe

Early in the evening of 3 December 1999, two squatters in an abandoned cold-storage warehouse in Worcester, Mass., knocked a lit candle onto a pile of clothing. At 6:13 p.m., an off-duty police officer reported smoke billowing from the roof. Within minutes, dozens of firefighters arrived on the scene.

At 6:26, a message reached the on-site commander that homeless people lived there, and he ordered crews to search the building. As smoke spread through the six-story maze of corridors and meat lockers, visibility began dropping fast.

At 6:47, a firefighter who had entered from the roof sent a radio transmission: “Rescue to Command, I need help on the floor below the top floor of the building. We are lost…”

Firefighters are specifically trained to deal with disorientation. In near-black conditions, they often use hoses or search ropes to find their way back to safety. Without tethers, they follow the “right-hand rule,” running the fingers of one hand along a wall until they reach a door or window through which they can escape.

The two lost firefighters didn’t have ropes, and their right hands presumably led them only deeper into the warehouse’s labyrinth. When they realized they were lost, they did what they’d been taught to do: They stopped moving, radioed for help, and activated the personal alert safety system (PASS) devices attached to their air tank harnesses. These battery-powered boxes emit a piercing alarm for rescuers to listen for.

None of the firefighters sent to find their colleagues heard the alarms. And because the warehouse was windowless, save for a few openings on the second floor, there was confusion about how many stories the building had. Not knowing which floor the lost men were on, rescuers wasted time searching the wrong ones.

As the fire raged, more firefighters were dispatched. At 7:10 p.m., a lieutenant made a distress call: “Chief, get a company up the stairwell to the fifth floor. We can’t locate the stairwell. Or give us some sign as to which way to go. We are running low on air and we want to get out of here.”

At 7:49, someone reported that the building might collapse. The heat and smoke were now nearly unbearable. When the commander finally called off the search at 8:00, six firefighters were missing.

Getting lost is a hazard firefighters accept with the job. They know that most times, someone will rescue them before their air tanks run out or flames engulf them. But they also know there’s a fair chance they will inhale smoke or get burned waiting. Each year in the United States, hundreds of firefighters suffer injuries, and a handful die because they can’t get out of burning buildings.

You might think there’s an easy tech fix. After all, satellite-based navigation systems do a fine job of maneuvering you through city streets or tracking down that laptop you left in a cab. But creating a system for navigating indoors, where satellite signals don’t easily penetrate, is much tougher. Yes, consumer products already exist for airports, malls, and museums. But most of these require prior surveys of Wi-Fi signals, which aren’t present everywhere, especially if access points or power sources go up in flames.

In the decade or so since the Worcester fire, dozens of engineers have joined the race to build an indoor navigation system that can track firefighters to within a meter. Now, two groups say they are close to succeeding. If they can clear just a few lingering technical hurdles, they may be able to get equipment in the hands of firefighters, paramedics, and other first responders within the next couple of years.

One of these groups was founded by John Orr, a professor of electrical and computer engineering at Worcester Polytechnic Institute (WPI). “I had the feeling that technology had to be able to solve this problem, where people were dying literally within 100 feet of safety,” he says.

He recruited fellow professors David Cyganski and Jim Duckworth to brainstorm solutions. “We thought it was going to be a two- or three-year project,” Duckworth says, flashing a rueful smile.

An early scheme the team devised called for three stationary radio transceivers that fire crews would set up outside a burning building. These would link wirelessly to one another and to a laptop-size base station, which would house an atomic clock and orient itself and the transceivers using GPS signals. The transceivers would listen for transmissions from small radios carried by the firefighters and pass those signals to the base station. By timing the signals’ arrival from each portable radio, the base station could determine their ranges and then calculate their three-dimensional positions using basic geometry. This location data would stream wirelessly to the incident commander’s computer display, which could be loaded with floor plans if any were available. Then if a firefighter got lost or injured, the commander could dictate an escape route or direct rescuers over voice radios. (No one expects firefighters in a crisis to deal with display screens themselves.)

The WPI team called its project the Precision Personnel Locator, or PPL. In theory, the system could achieve at least 1-meter precision. And indeed, early prototypes worked well in houses or small brick buildings. But they failed miserably in the presence of metal or concrete. Beams, joists, roofs, and rebar blocked or scrambled the radio waves, creating unmanageable multipath interference. “A commercial building is like a hall of mirrors,” Cyganski says.

For nearly two years, the WPI team chipped away at the project with only its department’s own funds. Then, on 11 September 2001, 412 firefighters and other emergency responders died in the aftermath of the attacks on the World Trade Center, in New York City. Suddenly, the ability to locate emergency workers became a national priority. So when the U.S. Department of Homeland Security formed its science and technology research arm in 2004, one of the new directorate’s first tasks was figuring out how to track people inside buildings.

Leading this effort was Jalal Mapar, who at the time was a program manager at the Homeland Security Advanced Research Projects Agency and now directs its resilient systems division. He soon reached out to the WPI group and began sponsoring annual indoor navigation workshops at the university. Although he praised the WPI group’s efforts, Mapar didn’t think an ideal system should include fixed equipment that took precious time to set up. He began funding private companies to come up with alternative systems.

“My basic requirement was no infrastructure,” says Mapar. “These things have to be self-contained and operate easily on their own.” He favored using a gadget known as an inertial measurement unit. An IMU uses gyroscopes and accelerometers to track a wearer’s movements in three dimensions. Modern microelectromechanical systems (MEMS) can squeeze these components into a matchbox-size package that firefighters can wear on their boots. Before a firefighter enters a building, this IMU-based locator would establish its initial position using a GPS receiver. It would then use a radio to beam position changes to the incident commander.

IMUs, however, are prone to drift. And because a unit determines its current position relative to its previous one, even tiny errors quickly add up. Just standing still can throw off many models, WPI’s Duckworth claims. “Within 2 minutes, it’ll look like you’ve drifted out of the building,” he says.

After several years of workshops at WPI, the experts all agreed that neither radio systems nor IMUs alone would work. By now, at least a dozen institutions and start-ups were trying to commercialize a product. Most of their prototypes used “fusion” algorithms to combine data from both radios and IMUs—and in some cases also magnetometers, pressure sensors, and digital compasses—to overcome the deficiencies of individual sensors. These systems performed well in the lab, and their inventors were itching to take them to the field.

In April 2009, Homeland Security hired the WPI group to assess six location systems that manufacturers thought were ready to market. (The department has kept the candidates’ names confidential, though Mapar says the list didn’t include the WPI team.)

At a police training academy in rural Massachusetts, more than 60 firefighters and police officers tested the systems in buildings they had never entered before. The firefighters blacked out their masks to simulate thick smoke, and the WPI engineers timed how long it took them to complete a series of rescues.

The results were abysmal. One system failed to detect sharp turns. Another directed a rescue team into the wrong building. Yet another worked when firefighters walked or ran but not when they crawled, which firefighters habitually do to avoid the worst of a fire’s smoke and heat. “It’s amazing how many [engineering] teams didn’t comprehend how firefighters went about their business,” Duckworth says.

None of the tech solutions helped firefighters complete missions faster than the old-school strategies, such as searching methodically and listening for PASS sirens. In some trials, rescuers took nearly twice as long to reach victims; in others, they never found them at all. To the engineers’ dismay, technology had made a bad problem worse.

With such disappointing commercial vendors, Mapar decided that Homeland Security would develop its own system, called Geospatial Location Accountability and Navigation System for Emergency Responders (GLANSER). In its current form, the system includes wearable units about the size of tissue boxes that attach to a firefighter’s breathing gear. Each device packages a military IMU with a GPS receiver, Doppler radars to correct velocity, a pressure sensor to judge changes in altitude, and a radio to measure the range to a fixed base station and to other nearby units. This radio network uses short pulses to help get around the multipath problem. Inside each unit, powerful algorithms known as Kalman filters combine all these inputs to make a precise prediction of its position.

GLANSER also enables up to 11 neighboring wearable units to communicate with the base station and with one another, forming a “mesh network.” Meshing lets the units swap location data, which improves individual accuracy even more. Two or three units working together can resolve their positions to at least 3 meters, Mapar says. In theory, adding more units would reduce that error, although Mapar will have to wait for the latest test results (due back this month) to know by how much.

Photo: Paul Martinka/AP Photo

Through meshing, units can also relay signals for one another. “If the signal from one unit can’t reach the base station, it’ll hop through the network to get out,” Mapar explains. GLANSER can support more firefighters by adding more base stations and hence more mesh networks.

During tests at WPI in August 2012, GLANSER became the first system to show an advantage over traditional methods. Using the system, rescuers navigated to victims without making a single wrong turn and even completed one mission in only 6 minutes, as opposed to the 30 minutes or so typically needed to make similar rescues without the technology. Now, a handful of U.S. fire departments are testing early models manufactured by Honeywell International.

The technology still faces some obstacles. One of the toughest challenges will be reducing the size and weight of the portable units so that they don’t impede firefighters’ movements. Because the battery makes up much of a unit’s bulk, power engineers at Homeland Security are working to develop a thin, flexible battery that could slip into the lining of a firefighter’s jacket. Just 1 square meter of the 1-millimeter-thick material, Mapar says, would be “the equivalent of 100 AA batteries.”

GLANSER’s biggest drawback, though, is its cost: The components of a portable unit alone add up to between US $2000 and $3000, not counting the fancy battery. A commercial product would sell for many times that. According to the U.S. National Fire Protection Association, nearly three-fourths of the country’s fire departments can’t even afford spark-safe radios at less than $750 a pop. Mapar suggests that cash-strapped departments could set up lease deals with GLANSER distributors, but even then the price may be prohibitive.

The WPI group, meanwhile, is addressing the cost problem head-on. “Our approach is constrained by making something that can be mass-produced and literally cost a few hundred dollars,” Duckworth says. To do this, he and his team aim to eliminate the need for expensive, power-hungry hardware by building better software.

In the group’s latest vision for the PPL, firefighters would carry portable units the size of walkie-talkies. Each would contain a cheap MEMS IMU and a pulsed radio transmitter to reduce multipath error. As in the original design, a base station would calculate the units’ positions using its atomic clock time and radio signals from three fixed transceivers. The WPI group envisions that the transceivers would be mounted on fire trucks or incorporated into ladders so that crews wouldn’t have to waste time setting them up on the ground.

In addition to transmitting ranging signals, each portable unit would broadcast its IMU’s internal position estimate. The base station would then fuse the two guesses into a more accurate one using a mathematical technique known as synthetic aperture imaging.

Military planes have long used this method to map ground terrain with radar. In traditional radar imaging, a physical antenna focuses an RF beam on a target and captures information from its reflections. Achieving higher resolutions requires narrower beams, which calls for larger antennas. To resolve fine details, such as tanks, a plane would need an antenna much longer than it could carry. Using synthetic aperture imaging, it can use a smaller antenna and simulate the larger version in software. In essence, the distance the physical antenna travels becomes the aperture of this virtual antenna.

From inside a fire truck, the PPL base station would do something similar using the transmissions from a firefighter’s portable unit. Just as the plane’s radar uses its flight path to virtually grow its antenna, custom-built algorithms at the base station would use the unit’s movements to create virtual antenna arrays. By evaluating the multiple input signals, the algorithms would pinpoint the firefighter’s location. Finally, the base station would feed these results back to the portable unit to help correct IMU drift.

The PPL has several advantages over GLANSER: The portable units are less cumbersome, they boast a 24-hour battery life, and they’ll likely be much cheaper to manufacture. Over the past couple of years, the Worcester Fire Department has helped assess the equipment during mock rescues. Although these tests haven’t been as rigorous as the evaluations of GLANSER, Duckworth says they’ve shown that the PPL can quickly locate victims, even in large steel or concrete buildings. The system is most accurate in residences, where errors are of no more than a meter, he says.

Still, the WPI group, like Mapar’s team, must overcome a few technical obstacles before its system is ready to market. If fewer than three engines respond to a fire, crews will need one or two ground-based transceivers, and the WPI team will need to find a way for firefighters to deploy these without delaying their mission. The engineers also need to make the portable units more rugged and optimize the synthetic aperture imaging algorithms, which were developed on powerful PCs, to run on embedded processors in a base station.

“We’re giving ourselves a difficult challenge,” Duckworth admits. “You have to get firefighters within 1 to 2 meters, and it has to be 100 percent reliable or they’ll never use it again.” He estimates that the PPL system could be ready for fire departments to test in as little as 18 months, provided his team can find a manufacturer to license it.

It took eight days to uncover the bodies of the six firefighters lost during the Worcester warehouse blaze. Investigators later determined that the squatters probably escaped the building before the first engine even arrived.

Today a fire station stands where the warehouse burned down. A bronze sculpture of a kneeling firefighter, a plaque, and a portrait of six teammates carved in a granite wall commemorate the fallen men. If you drive by any day of the week, you will likely see crews cleaning trucks, testing gear, or running drills. These men and women willingly put themselves in harm’s way to save others’ lives. It’s high time they had the tools to save their own.

This article originally appeared in print as “The Way Through the Flames.”

About the Author

Mark Harris, a contributing editor, won a Grand Neal Award for his article “A Shocking Truth,” which appeared in the March 2012 issue of Spectrum. As a freelance writer and Knight Science Journalism Fellow, he prides himself on sniffing out riveting stories. When he learned that a tragic warehouse fire in a small Massachusetts town had ignited a race to build an indoor navigation system that could keep firefighters safe, he says, he just had find out how far the technology had come.