Threat Evaluation Weapon Assignment
In fighter planes, identifying the enemy’s behavior allows in some cases to predict its future trajectory. Two steps are necessary:
- The identification of maneuvers is done from the radar track associated with the enemy aircraft: HMMs (Hidden Markov Models) are Bayesian methods that allow to filter trajectories according to their type (attack, crank, retreat …) and finally to associate a probability to each type of behavior.
- The generation of trajectory bundles takes into account the uncertainty on the dynamic capabilities of the platform related to its type. Probabilistic methods such as Gaussian processes allow to draw probable trajectories of the aircraft and to identify the probability distribution on its position after a given time. This ability allows to anticipate the launch of a countermeasure
These two methods are based on Monte-Carlo draws made on a large number of variables.
The computations must occur in real time in embedded devices, which would highly benefit from specific hardware acceleration.