Melbourne-based Sentient Vision Systems plans to adapt its unique Artificial Intelligence (AI) and Machine Learning technology to Short-Wave Infrared (SWIR). The program is designed to provide the company’s proven stand-off ViDAR (for Visual Detection and Ranging) optical radar capability in the most challenging optical environments.
Haze – basically, smoke, dust or moisture particles suspended in the air, or all three – reduces visibility, and defies attempts to see through it using the Electro-Optic and Infrared (EO/IR) sensors that are in common use aboard airborne platforms over the sea and inland.
The company will blend a SWIR sensor with AI derived from its proven ViDAR system to create a sensor imagery analysis capability that has never existed before. ViDAR is a software-based imagery analysis system that examines every frame in a sensor’s high-resolution imagery feed and detects targets that would be invisible to a human operator, or very hard to spot. It offers up to 96% probability of detecting a target and enables a patrol or SAR aircraft to cover a designated search area up to 30 times faster than one without a ViDAR sensor.
A SWIR sensor operates in the 1-3 micron waveband which gives it a unique ability to ‘see’ through atmospheric haze, whether moisture, dust or smoke. By contrast, Medium-Wave IR (MWIR) in the 3-5 micron waveband is well suited to night vision and poor visibility and is used by most ViDAR operators. Long-Wave IR (LWIR) in the 8-15 micron waveband is the traditional ‘thermal imaging’ wave band and LWIR sensors detect temperature differences with great sensitivity, but are attenuated by haze, dust and smoke.
Whereas a conventional EO/IR sensor might have a range of up to 20nm in good atmospheric conditions, in a hot, humid and hazy environment its range could fall to as little as 1nm in thick haze, and even less than this in dense smoke or dust.
Using SWIR sensors this range increases considerably, and the ViDAR-derived AI and Machine Learning technology means that difficult to spot targets within the sensor feed can be detected autonomously in exactly the same way as a conventional, MWIR ViDAR system. ViDAR simply puts a thumbnail on the operator’s screen that provides the range and bearing of the target for closer investigation by the platform’s primary sensor.
This enhanced detection capability offers significant advantages in Search and Rescue (SAR) missions, aerial firefighting, Law Enforcement (LE) and maritime surveillance.
The advantages of SWIR and ViDAR to the operator are clear: it’s a passive capability, so undetectable, unlike radar. A SWIR sensor provides an EO-like image of the target, which is much easier to interpret than a radar image. The combination of SWIR and ViDAR is especially good for detecting small objects, unlike radar: ViDAR’s combination of AI and Machine Learning enables the sensor to detect and classify targets smaller than 1 pixel in size. So, a SWIR/ViDAR-equipped aircraft or UAV is far more productive than an aircraft without this system, says Sentient Vision Systems. However Sentient Vision Systems says this capability will augment, not replace, its existing, proven MWIR ViDAR system.
Sentient Vision Systems is still exploring the technology and defining the right architecture for the system, but early indications are good. With the worst effects of the COVID-19 pandemic expected to pass soon, the company plans to verify and validate the SWIR/ViDAR combination in real-world conditions this year.