Velocity-Dependent Orientation Estimation Using Variance Adaptation for Extended Object Tracking

For extended object tracking (EOT), the shape of an extended object (EO) is usually fixed (e.g., tracking vehicles), but the orientation varies. Thus, an accurate estimate of the time-varying orientation is important. The orientation and the heading are not always identical but highly dependent, which can be used as additional prior information to improve estimation accuracy, including the orientation. In view of this, this letter proposes a velocity-dependent orientation estimation approach to EOT utilizing this information. First, we model the quantity between the orientation and the heading as a Gaussian noise with zero mean and adaptive variance. Second, based on the proposed model and the integration of a pseudo-measurement, a variational Bayesian (VB) approach is proposed to estimate the kinematic, shape, and orientation states. The proposed approach can adapt to most dynamic scenarios without the need for a sophisticated mathematical model. The effectiveness of the proposed model and estimation approach is demonstrated by using simulated data.