Sensor data association using relative positions among targets and bias estimation between separate sensors

Sensor data association is an important problem in modern multi-sensor systems. The purpose to solve this problem is to decide which measurements from the different sensors belong to the same target. The traditional methods for data association usually form the association matrix and find the optimal solution in different ways. These solutions are, however, sensitive to the characteristics of the sensors. A novel method which uses the relative positions among the targets and extracts the relative positions pattern to compare and search for the matching pairs between the separate sensor systems is proposed. An improved algorithm which is suitable for sensor data association using relative positions is also presented. The sensor bias has little influence upon this association algorithm due to the inherent characteristics of the relative positions. Simulation results show that association using relative positions is robust against sensor bias and has an overall improvement.