Pirates don’t expect to have to fight AI as well

Published on: August 2020

There’s an entire genre of fiction, and some actual history, devoted to piracy and illegal activity on the high seas. Pirates have survived to this day because it’s so hard to first spot them and then stop them.

The challenges facing today’s Naval and para-military forces, such as Coastguards and maritime police, haven’t changed since the 17th and 18th centuries. They need to establish situational awareness and a Local Operating Picture (LOP). They need to be able to spot suspicious activity and investigate it quickly; and they need to be able to respond effectively to what they find – mariners in distress or, more commonly, some kind of illegal activity that needs a timely, evidentiary quality record that can be put before a court.

Some operators confront mischief-making by a non-state actor or sovereign power. The appropriate and proportionate response might be anything from a censorious, escorting presence to an anti-ship missile – a Navy needs to be able to do both, and anything in between. This is the thinking behind the Royal Australian Navy’s Project SEA 129 Ph.5 and Maritime UAV Continuous Development Program. They are designed to enhance the Navy’s airborne Intelligence, Surveillance and Reconnaissance (ISR) and Targeting capabilities to help it deliver whatever effect the situation demands. The RAN is typical of modern maritime forces in exploring this elegant and economical response to common maritime security challenges.

Melbourne-based Sentient Vision Systems is helping its clients (including the RAN) build surveillance architectures with capabilities that deliver that ISR capability. An outstanding sensor suite to detect and then prosecute the smallest and most uncooperative of targets is essential, but it’s not enough. The advantage comes from Artificial Intelligence (AI) and deep learning that automatically enable mission systems and their operators to create a LOP and the essential context, history and pattern of activity that makes it meaningful.

Sentient Vision Systems manufactures ViDAR, the world’s first optical radar. An acronym for Visual Detection and Ranging, ViDAR consists of multiple fixed, high resolution cameras with a combined Field of View of 180 degrees. Mounted in an external pod or internally aboard an aircraft or UAV, ViDAR can passively find an object in the water as small as a human head.

Sentient Vision Systems has integrated ViDAR with multiple different airborne mission systems and platforms ranging from a UAV such as the Insitu Scan Eagle® and UMS Skeldar V-200, to the Beechcraft Kingair 350 and Viking Twin Otter fixed-wing aircraft, right up to the Bombardier Challenger 604 jet. The ViDAR sensor payload is used for maritime Search and Rescue (SAR) as well as constabulary and counter-terrorism operations.

Maritime Patrol missions generally focus on illegal activity: people and narcotics smuggling, Illegal Unreported and Unregulated fishing (IUU) and terrorist operations. Many of their targets are diminutive wooden or GRP craft with a small or non-existent radar cross-section. And often maritime authorities need to detect potential targets from a safe or discreet distance.

If the target is within its field of view, ViDAR will spot it, even in Sea State 6 – it has a proven Probability of Detection of 96%, thanks to its AI enabled signal processing software. ViDAR can detect commercial vessels at ranges of more than 30nm, or pleasure craft at more than 10nm, then provides a thumbnail to the main screen on the operator’s mission system showing the target and its location. This enables the operator to cross-cue the aircraft’s primary EO/IR sensor towards it for inspection and generation of high-quality imagery and location information for evidentiary and intelligence purposes ViDAR’s speed and performance dramatically increases the coverage area by more than 300 times compared to traditional visual search techniques.

ViDAR is in service with many customers worldwide, including the RAN and US Coast Guard, but Sentient Vision Systems points out that it forms part of an integrated airborne surveillance architecture. Its AI-enabled signal processing capability enables it to detect miniscule targets against an extremely cluttered background. Data fusion and machine learning over multiple missions enables faster, less-fatiguing aerial searches in all sea conditions and a quicker response to maritime threats and emergencies.

Once detected, that AI helps classify and sort targets so they can be built into the LOP – a comprehensive real-time picture of activity across a sea area that’s built up from ViDAR itself, surveillance radar and returns from ships’ Automatic Identification System (AIS) transponders. And AI helps collate simultaneous inputs from ViDAR and other sensors mounted on multiple manned and unmanned platforms.

Regularly re-visiting an area enables Pattern Recognition – establishing the normal level and type of activity in a specific sea area and then spotting anomalies. That ‘normality’ might be the types of craft in the area – fishing fleets, regular ferry services – and an anomaly could be anything from a vessel of unusual size or that’s travelling in a counter-intuitive direction, to one that’s deliberately hard to spot. Enabled by AI, and refined over multiple missions by ViDAR’s machine learning, this is a powerful surveillance capability.

This is what Sentient Vision Systems is helping customers achieve. The company is investing heavily in AI and the signal and image processing capabilities of ViDAR and other detection products such as Kestrel. This ‘secret sauce’ enables a rapid, and if necessary discreet, search for uncooperative targets, the creation of an accurate LOP and an ongoing comparison with previous patterns of behaviour in that same sea area.

And once a target is located and positively identified, the appropriate action follows: a suspicious vessel might be shadowed discreetly using a UAV; it may be followed and boarded, from a helicopter or RIB; a distressed mariner may be assisted by a manned helicopter or boat; or an airborne EO/IR and radar sensor suite may provide over-the-horizon targeting for an anti-ship missile.

These outcomes are all enabled by Sentient Vision System’s mastery of sensor design, signal processing and AI: a quicker search, quicker target identification and quicker prosecution.


Click here to download