The intent inference in target tracking based on the target track

Abstract

The detection of anomalous trajectory patterns in target tracking is considered here to infer the target's intent which is assumed to be related to the shape of the target's trajectory. The context-free grammar is used to model the trajectory of the target and obtain the mode sequence via the quantification to the state estimate. The modified Earley algorithm is proposed to parse the mode sequence in real time and infer the intent of a target under conditions on the target identification. Numerical simulation examples show the validation of the grammar to model the rectangle trajectory with three turns at any moments.