ION GNSS 2010
Session A2: Next Generation GNSS Integrity for Aviation

Title: Ellipsoidal Set Based Filter for Bearing-only Maneuvering Target Tracking
Author(s): Y. Liu and Y. Zhao, Beihang University, China

In most solutions to maneuvering target tracking, the state transition and measurement models are assumed a prior statistical description of the maneuver, ranging from white noise to colored noise to a semi-Markov process. However, those algorithms work well within the context of statistical assumptions of maneuver process. But if any prior knowledge about the maneuver target is available, algorithms performance may be degraded. And the actual nature of the maneuvers are not corresponded to any statistical assumptions. In this paper, we consider a two-state maneuvering tracking problem with bearing-only maneuvering target tracking using two static platforms. For maneuvering target tracking, the state transition and bearing-only measurement process are nonlinear and modeled as non-stochastic processes with unknown but bounded noise terms. The approach does not rely a statistical description of the maneuver as a random process. In these scenarios, ellipsoidal set based estimation provides an alternative approach to the statistical estimators. The bearing measurement with bounded noise defines a set in position space which must contain the feasible target state. Important for this work, the state transition noise and measurement process noise are functions of the system´s state. A detection scheme has been developed to determine that a maneuver is indeed occurring. When the target is maneuvering, we consider the impact of maneuvering target as an increased bounded noise in bearing measurement process. The ellipsoidal set based framework assumes two known bounded for both state transition and bearing maneuvering process. And the initial target state is assumed in a known bound. We use typical approach to choose the optimal volume ellipsoids to overcast the set of possible states in both time and measurement updates. The ellipsoidal set based bearings only filter has been compared to the extended Kalman filter in a Monte Carlo simulation. It is shown that the filter is successful in accommodating an unknown nonlinear model for a maneuvering target tracking scenario.

Previous Abstract Return to Session A2 Next Abstract