To improve the passive tracking situation of receiver for drogue in AAR’s “capture” phase, predictive guidance scheme was implemented. With help of AR model, predicted drogue position was regarded as target point of receiver, and parameters of the model were estimated by Kalman filter principle. Phenomena throughout the AAR simulation experiment were analyzed quantitatively, and four optimization measures were proposed to improve real-time performance of the scheme. The simulation results indicate that proposed algorithm possesses high prediction precision and that optimized scheme significantly enhance success rate of AAR, which is of crucial importance in achievement of AAR technique.