Joint Tracking and Identification Based on Constrained Joint Decision and Estimation

As a critical component of traffic safety, violation detection and treatment can effectively reduce illegal driving so as to ensure the safety of our lives. In this system, both road-constrained target tracking and identification are involved, which are coupled. This is a constrained joint tracking and identification (CJTI) problem, and good solutions require addressing the two problems jointly. The recently proposed joint decision and estimation (JDE) framework and the conditional JDE (CJDE) approach have been proved to be superior for such problems involving coupled decision (identification) and estimation (tracking). Inspired by this, for CJTI of ground targets, this paper proposes a novel constrained CJDE approach. First, a hybrid CJTI system incorporating target dynamics, class characteristics and road constraints is formulated. Based on this, we propose a constrained CJDE risk converting a constrained tracking cost and a constrained identification risk into one framework. An optimal constrained CJDE solution containing an estimator and a decider is derived. Simulation results verify the effectiveness of the proposed constrained CJDE method. They show that by fully utilizing the constraint information and the coupling between tracking and identification, the constrained CJDE approach outperforms traditional two-step methods in joint performance.