A new approach in distributed multisensor tracking systems based on Kalman filter methods

In multisensor tracking systems, the state fusion also known as ”track to track” fusion is a crucial issue where the derivation of the ”best” track combination is obtained according to a stochastic criteria in a minimum variance sense. Recently, sub-optimal weighted combination fusion algorithms involving matrices and scalars were developed. However, hence they only depend on the initial parameters of the system motion model and noise characteristics, these techniques are not robust against erroneous measures and unstable environment. To overcome this drawbacks, this work introduces a new approach to the optimal decentralized state fusion that copes with erroneous observations and system shortcomings. The simulations results show the effectiveness of the proposed approach. Moreover, the reduced complexity of the designed algorithm is well suited for real-time implementation.