Multiple Target Tracking with Multiple Finite Resolution Sensors

In this paper tracking multiple targets with multiple finite-resolution sensors is presented with emphasis on the measurement-to-track association with possibly unresolved measurements. Two data association algorithms are considered. The multiple scan data association problem is first formulated as a multidimensional assignment problem with explicit new constraints for the unresolved measurements. The auction algorithm for 2-D assignment with unresolved measurements is discussed and the multistage Lagrangian relaxation approach for solving the N-D assignment problem (for an N−1 frame sliding window) is also generalized to the current finite-resolution sensor situation. Subsequently, a modification to the assignment problem is developed that leads to a linear program (LP) where the optimal solution can be a noninteger in [0, 1]. The fractional optimal solution is interpreted as data association (pseudo)probabilities over the N − 1 frame sliding window. Consequently, the target states are updated by a probabilistic data association (PDA) filter. The hard (binary) decision assignment solution is compared with the soft decision (LP) solution for a twodimensional tracking problem of two closely-spaced targets with various sensor configurations. Based on the simulation results, the soft assignment approach has better track maintenance capability while its computational load is somewhat higher than that of the hard assignment approach.