A nonlinear robust model predictive differential game guidance algorithm based on the particle swarm optimization

Two-dimensional engagement of a pursuer and a maneuvering target, affected by matched uncertainties, is formulated as a nonlinear differential game. The uncertain guidance problem is converted into a nonlinear model predictive control problem by introducing an appropriate cost function. The objective is to calculate the best guidance commands of the pursuer and the worst possible target maneuvers simultaneously, over a receding horizon. The proposed cost function penalizes the line-of-sight rate, the pursuer acceleration, and the uncertainties. It also rewards the target maneuver. A particle swarm-based dynamic optimization algorithm is developed to solve the nonlinear model predictive differential game, affected by the uncertainties. Performance of the proposed guidance algorithm is evaluated against maneuvering and non-maneuvering targets. The algorithm is also evaluated for the cases when the pursuer has a high initial heading error, and the guidance command is constrained. The statistical performance of the proposed algorithm is evaluated using Monte Carlo simulation. Moreover, a processor in the loop experiment is performed to verify the implementation capability of the proposed algorithm. Finally, a comparison is made between the performance of the suggested algorithm with some other methods including linear- quadratic differential game, state-dependent Riccati equation-differential game, a guidance law on the basis of adaptive dynamic programming, a proportional navigation guidance improved by particle swarm optimization, a guidance algorithm based on the continuous ant colony controller, switched bias proportional navigation, and augmented proportional navigation.