State Estimation With Heading Constraints for On-Road Vehicle Tracking

In transportation networks, vehicle motions are usually subject to the constraints arising from the roads, preset sea-routes, or preset flight routes. Taking advantage of prior known information of such constraints generally produces better surveillance performance. The problem of state estimation with heading constraints, in which only the direction of the target trajectory is prior known, is considered. When only part of the trajectory information is available, the conventional constrained estimation methods cannot be used to produce constrained estimates. The heading constraints in two typical situations are investigated, one is with a straight line, the other is a circular arc. For the straight trajectory, two augmentation approaches are proposed to formulate the constraints. One, parameter augmentation, augments the base state by the
$y$
-intercept of the constraint straight line. In the other approach, state augmentation, the states at the past time step are used to augment the base state. Two corresponding heading constraints are formulated by the augmented state’s elements and the prior known information about road direction. For a circular trajectory, the heading constraint is formulated using position and velocity components in the base state. Pseudo-measurements are constructed to incorporate the heading constraint into the estimators and result in three heading constraint Kalman filters (HCKFs). They are parameter augmentation HCKF and state augmentation HCKF in case of straight trajectory, and HCKF for circular trajectory. The discrimination of road segments at junctions is also discussed. Furthermore, the proposed constraint filters are integrated into the interacting multiple model estimator to handle on-road maneuvers with possible accelerations. Numerical simulations are conducted to evaluate the performances of the three HCKFs compared with existing methods.