Maximizing kill probability using bayesian decision-directed guidance

Guidance laws are commonly designed to minimize the miss distance. Then, given a particular miss-distance-minimizing guidance law, thorough Monte Carlo simulation studies are performed to determine the required lethality radius for the warhead, assuming nominal targets in nominal stochastic scenarios. However, this approach comes with a caveat: when a nonnominal situation is encountered, the designed warhead might fail to ensure the capture of the target with a sufficiently high probability, leading to degradation in the interceptor s guaranteed performance. In the present, work we propose an inverse paradigm whereby, for a given warhead (with a given lethality model), a guidance law is designed to maximize the interceptor s kill probability against any target. As opposed to the standard (hard) approach, which models the warhead lethality as an indicator function of the miss distance, we use a more realistic, soft, probabilistic model. We base the solution on a deterministic differential-game-based guidance law and modify it to incorporate the probabilistic lethality model and the imperfect information about the target state using Bayesian decision theory. We demonstrate the performance improvement associated with the new approach via an extensive Monte Carlo simulation study involving several interceptors with various warheads.