The Army is going out to the cutting edge in the push to uncover buried bombs. It’s leveraging artificial intelligence and augmented reality in its latest countermine efforts.

“We want to reduce the training burden, to improve the soldier-machine interface, to use machine learning to improve performance,” said Michael Grove, principal deputy to the director of the Night Vision and Electronic Sensors Directorate at CERDEC, the Army’s Communications-Electronics Research, Development and Engineering Center.

To that end, the Army is presently in testing and development of a new technology suite known as real-time spatial location tracking.

Present sensor systems for dismounted soldiers are effective but have some serious operational constraints. They can be complicated to use and have a tendency toward false positives.

“With sensor technology you are limited by signal-to-noise or signal-to-clutter: things that aren’t the target but can look like a target. That can be rocks or roots or fragments of metal. Those things make it very difficult to pull out that signal,” Grove said.

At the same time, Army is feeling a high sense of urgency around the need to detect mines in a way that is consistent, accurate and easy to use.

“Many of our adversaries still maintain large stockpiles of landmines. There are large minefields currently in potential conflict zones including Korea and the Ukraine,” Grove said. “We also can’t forget about IEDs [improvised explosive devices], which caused over 36,000 casualties in Iraq and Afghanistan.”

The emerging solution would allow soldiers to sweep a sensor over the ground and get a quick, reliable visual readout on anything that may be lurking below the surface. It will rely on machine learning and AI to differentiate harmless matter from potentially deadly devices.

Emerging solution

Real-time spatial location uses stereoscopic sensors to create a visual map of objects under the ground, at the same time converting that visual read into digital data that can be interpreted by machine leading systems. AI applies pattern recognition capabilities to determine the exact nature of the suspect material.

“That allows us to do automated target recognition,” Grove said, “Instead of just saying, ‘I have heard something like this before,’ you now get an actual display mapping out the dimensions of the device. You can then invoke the AI as a highly efficient way to say ‘Yes, that is a mine of this type.’ And because the data is digitized you can send it off and use it to train other algorithms to improve their performance.”

The stereoscopic sensors are important because they allow the soldier to get a more precise read on the location of the potential mine.

“You want a good correlation between the image and the actual location on the ground. You want to make sure you know exactly where your target is, and not six inches off or 12 inches off,” Grove said. “The stereoscopic image gives you a way to achieve that level of location and navigation.”

The system employs augmented reality in the sense that digital representations are layered over a real-world view. A soldier still sees the ground, but it’s lit up in color, with orange representing a potential mine or other metallic object.

The system can likewise layer historical data on top of the real-world visuals. “Was there an explosive hazard detected in this area in the last five days? If you can give an individual a heads-up that a device was found here or that there was enemy activity in this area, that can enhance the performance of the detection system,” Grove said.

The push for high-tech countermine solutions is part of a larger effort to leverage advanced technologies to deal with buried explosives. In mid-2017, for instance, Army introduced the Standoff Robotic Explosive Hazard Detection System (SREHD), which uses robotic support to mark out potential hazards on the ground.

“In the future, we need to embrace and leverage robotics and understand it can be used to help us,” said Maj. Lendrick James, assistant product manager, in an Army news release.

Grove’s team has been collaborating with the SREHD developers with an eye toward eventually combining the AI advances and the robotics approach. “We share lessons learned back and forth. Some of the testing done under SREHD clearly will feed future developments on the handheld side,” he said.

The exact soldier interface of the real-time spatial location system is still under development by the CERDEC team in collaboration with the Maneuver Support Center of Excellence at Fort Leonard Wood, Missouri. They’re considering three possible formats: a wand-mounted display that connects back to the Nett Warrior suite; a see-through heads-up display; or possibly a monocle.

The final form will likely be determined by mission requirements. “It will depend on what the soldiers like,” Grove said.

Share:
More In Sensors