This post is also available in: heעברית (Hebrew)

In recent years, an evolution has been taking place at the arena of military detection known as spectral sensing, with software-driven approaches emerging.

Spectral sensing (also called hyperspectral) uses reflected light to determine the material properties of an object. It’s been used for decades in aerial implementations as a means to detect troops and weaponry. Recent evolutions have made spectral sensing a viable tool for ground troops, according to

“With an advanced set of algorithms we can tease out a lot of information from that spectral data. The hardware is still important, but it is what you do with that data that is absolutely critical,” said Dave Caudle, business development executive with ChemImage Sensor Systems.

“What has sparked the change in the industry has been the two conflicts in Iraq and Afghanistan, where America was trying very hard to solve the IED and VBIED [vehicle-borne improvised explosive device] problems,” Caudle said. These types of attacks have driven an urgent need for a more capable standoff-sensing capability.  

“Hardware solutions might pick up a military-grade explosive ingredient or a homemade ingredient, but we really needed an adaptable sensor so that if the enemy stopped using one ingredient and started using another, the sensors would pick up on that change,” he said.

This is where the power of software has come into play. In conventional spectral sensors, operators might need to swap out filters, sensors and lens in order to search for different elements. Algorithm-driven devices can rotate through a full library of potential hazards without the operator needing to make any adjustments.

ChemImage isn’t alone in this effort. In October, the Army’s Project Manager Soldier Weapons (PM SW) program demonstrated technology from Surface Optics Corp. Other Army programs have shown similar interest.

ChemImage has developed a shoebox-sized response to the IED and VBIED threat. The devices weigh eight to 10 pounds and can be configured to automatically scan for a range of potential hazards. Such a handheld sensor can help to keep soldiers a safe distance from any possible threat.

In the absence of spectral analysis, a soldier at a checkpoint has to question drivers and manually search vehicles. Thanks to improved algorithms, hyperspectral imaging “has become mature enough to allow that same service member to detect threats with a high degree of confidence from a safe standoff distance,” he said.

In addition to detecting explosives, such a system could be set to look for “precursors” – the ingredients that might be used to create homemade explosives. Sensors can pick up biological or nuclear contamination as well. Forensic uses also emerge.

While software-driven approaches have extended the range and versatility of spectral analysis, the technology has one significant limitation. Sensors analyze reflected light, so there needs to be a strong, reliable light source present. ChemImage addresses this by integrating a flashlight into its devices.

“In the places where these sensors are deployed, there are harsh environments. There are sandstorms, there are changing conditions. So keeping consistent light on the target has to be a priority,” Caudle said.