Intent Inference and Detection of Anomalous Trajectories: A Metalevel Tracking Approach

Classical target tracking assumes a state‐space model with target maneuvers modeled as a finite state Markov chain. This chapter motivates by metalevel target tracking applications on longer time scales. It develops metalevel tracking algorithms that assist human operators by extracting spatial patterns from target tracks to identify suspicious/anomalous spatial trajectories. The targets' spatial trajectories are modeled by two types of random processes: stochastic context‐free grammars (SCFGs); and reciprocal random processes. The chapter presents Bayesian filtering and smoothing algorithms for these processes to estimate the trajectories and hence intent of the targets. It describes a system‐level description of the anomalous trajectory classification problem, discusses the radar tracking framework and provides a mathematical description of trajectory classification. The chapter presents SCFG models and associated signal processing algorithms for trajectory modeling and classification. It considers reciprocal processes for modeling target intent. The parsing of the motion trajectories is implemented with Earley‐Stolcke parsing algorithm.