When GPS goes missing, infantry doesn’t miss a step
In the past year, visitors near certain military facilities might have noticed Soldiers and Marines with body-mounted cameras performing strange exercises, including spinning in place, rolling on the ground, crab walking, shuffling backward, and leaping back and forth.
These warfighters weren’t training for a performing arts program. Instead, they were helping to train an artificial intelligence program — one Leidos has developed as part of a system that ensures service members on foot always know exactly where they are, even when GPS can’t tell them.
Phone-based GPS has been an enormous boon to pedestrians of all sorts, whether it’s a tourist wandering through an unfamiliar city, or a backwoods hiker miles from an established trail. But the resulting dependency on GPS to stay on track can become a liability when the signal weakens or even drops out, as it often does near buildings or mountain ranges, inside tunnels and caverns, or under dense foliage.
Infantry fighters have become just as dependent on GPS as civilians, but the stakes are much higher. “They need to know exactly where they are and where friendly forces are located to complete missions on time, avoid accidentally getting into dangerous situations, and call in support,” says Jonathan Wetherbee, Leidos program manager for DARPA’s Squad X Core Technologies navigation program. Magnifying these risks, Wetherbee adds, is the fact that opposing forces may attempt to purposely disable or corrupt GPS signals.
An alternative to GPS
When the GPS goes out for warfighters on foot, Leidos Assured Data Engine for Positioning and Timing (ADEPT) kicks in, providing near-GPS quality location and navigation data. ADEPT was originally developed as part of the All Source Positioning and Navigation (ASPN) DARPA program. It can pull an accurate fix from several types of sensor data typically available to dismounted infantry, and it can run on a handheld device that clips to a belt. ADEPT can even be added to an existing tactical phone or radio.
A starting point for an ADEPT location fix is the chip-mounted gyroscopes and accelerometers found in military and even most civilian smartphones. These inertial measurement devices (IMUs) can sense motion, and keep track of how its location changes. But the tiny, inexpensive versions found in handheld gadgets aren’t very accurate without frequent GPS updates, and over even short periods of time can lead to significant errors in location estimates.
To significantly bound the errors caused by inertial drift, ADEPT can draw on other inputs, says Jordan Britt, a Leidos navigation subject matter expert on the ADEPT development team. “We have unique algorithms that can leverage measurements from a number of different sensors,” Britt explains. These can include camera images from body-mounted cameras, electronic compass readings, and altitudes from barometric altimeters — all commonly at hand for Soldiers and Marines on foot. ADEPT can even make use of radio signals picked up by handheld devices, notes Britt, including those from cellphone towers, Wi-Fi networks, and various types of satellites.
All the available sensor data is continuously monitored and processed by ADEPT’s “sensor fusion engine” software, which uses statistical metrics to figure out which signals are most trustworthy and how to best combine and weight them to get the most reliable position fix. For example, compass readings can be combined with inertial data to get a better estimate of motion, and altimeter data can be combined with camera images to try to match the terrain and landmarks to maps. The result is that without the benefit of any external signals, ADEPT can keep on-foot personnel updated on their location within tens of meters even after hiking for miles.
There were plenty of challenges to getting ADEPT to work with dismounted personnel. For one thing, the need to keep size, weight, and power (SWaP) of the gear to a minimum eliminates the sort of sophisticated sensor devices that can be placed in Humvees, tanks, aircraft, and boats. Even worse, most vehicles tend to move in relatively smooth and predictable ways — generally forward, with occasional turns. Infantry members, in contrast, routinely shuffle sideways to avoid obstacles, walk backward to keep threats in view, crawl on the ground to stay behind cover, and may even jump, spin and roll when the action gets intense. Hence the need to get Soldiers and Marines performing those strange maneuvers in order to train the system. “The algorithms had to learn to recognize these behaviors and the confusing sensor data they generate, to build up a better model of how people are moving,” explains Wetherbee.
To even more precisely nail down positions, ADEPT can weave together the sensor data from multiple personnel in the area — an entire squad on a mission, for example — even if individuals are as far apart as 300 meters. If just one person in the squad gets a good fix, the system can give others in the squad collaborative fixes that are nearly as good by accessing the ranging data between squad members that are often provided by tactical radios. And if any vehicles can be ranged, their navigation capabilities can be distributed throughout a dismounted squad as well. “Any good location information can be leveraged throughout a group,” says Britt.
That ability to get good fixes when GPS is denied means Soldiers and Marines can make the most of the standard tactics that have been built around having that data. And since opposing forces aren’t likely to have that same GPS-denied navigation capability, there’s even an opportunity for new tactics that take advantage of that edge. “Our personnel will be freer to operate out of line-of-sight on missions,” offers Wetherbee as an example.
In other words, the loss of a GPS signal could be something that infantry members stop worrying about, and maybe even start hoping for.