A Complementary Technology
Target classification is an important part of airspace surveillance. Whether you are monitoring with one sensor or a networked suite, without understanding what is being detected by those sensors it is nearly impossible to build up full situational awareness.
There are a number of ways in which detected targets can be classified; these are usually built-in to individual sensor units, and dependent on the technology of that sensor. For example:
Camera units can have a classification AI (using computer vision) built-in or connected to them
Radar sensors can use a technique called micro-Doppler to perform classification; this essentially works by detecting repeating motions (such as wing beats or rotor movements)
Dynamic Intelligence Solutions’ Octa technology represents a new way to classify targets. Rather than utilising the appearance of an object or a sensors’ specific response to it, Octa classifies objects based on “track dynamics”. This is determining the type of object based on the way it flies, using its location history - or track - as input. Our AI-enabled system makes these predictions using a combination of flight characteristics, extracted from the raw track data following an advanced statistical filtering process.
Because Octa works with the raw track data, it is sensor-agnostic; as long as a sensor can provide a series of location updates for a target, Octa can predict the type of object it belongs to. This extends to multi-sensor systems too; Octa can process track feeds from multiple sensors simultaneously, and works with correlated targets (where multiple sensors detect the same object) in exactly the same way as others.
All of this means that Octa is a complementary technology. Instead of replacing existing capability - or requiring additional capital investment in hardware - it works alongside it to either add classification features to sensors, or improve existing classifications by acting as a “second opinion”. This helps systems to operate:
At scale: Octa can help systems operate at scale, by analysing many targets simultaneously. This means it can help to prioritise the work of sensors which don’t scale as well, such as cameras.
At range: Octa can help systems classify further out than previously possible. For example, most micro-Doppler systems can detect objects at a much greater distance than they are able to classify. Adding Octa means that those sensors can make classification determinations out to a larger distance than they otherwise could.
Adding a complementary classification capability like Octa can supercharge existing sensors without the need to invest in additional hardware deployments. By fusing methods and techniques, a holistic, single operating picture can be built up, maximising advantage whilst minimising the disadvantages of individual detection systems.