Maneuver Recognition Based on Affine Weighted Dynamic Time Warping Algorithm

Abstract

For same type of maneuvers, parameters of maneuver sequences may vary greatly due to different lengths of time, and parameters of maneuver sequences are more random, which cause the accurate identification of maneuver types still have some challenges. In this paper, a typical tactical maneuver database is established based on the analysis of relationships between tactical maneuvers and combat intent, and it is used as a tactical maneuver template. Moreover, an affine weighted dynamic time warping (AWDTW) algorithm is proposed for maneuver identification, which has three advantages: (1) Influences on maneuver recognition are considered for different dimensions parameters of maneuver sequences. (2) Affine transformations are introduced to solve the problems on scaling and offset of maneuver sequence parameters due to different lengths of time. (3) Real-time of the proposed algorithm is improved by limiting path search intervals in the tactical maneuver template. Finally, simulation analyses are performed on maneuver sequences in different maneuver spaces, simulation results show that the AWDTW algorithm achieves effective identification of tactical maneuver types.