Abstract:
As the first step of data association, the association gate is the prerequisite for ensuring that the measurement traces and the target state estimation can be correctly associated. For a multi-target system under clutter environment, a small association gate may cause the actual measurement of target to fall outside the gate and result in the loss of the target. On the other hand, larger association gate may bring too many irrelevant measurement, which might not only increase the computational complexity of the joint probabilistic data association algorithm, but also affect the tracking accuracy. Therefore, in order to minimize the interference of irrelevant measurements and improve the probability of correct association, it is necessary to properly control the association gate according to different correlation situations. To this end, this paper proposes a novel adaptive association gate algorithm to overcome the shortcoming of the traditional association gate design method that easily causes wrong target tracking and low tracking accuracy in multi-target tracking. This proposed algorithm is based on an association performance evaluation, which is used to evaluate the difference between current state estimation and measurement values such that the performance of association module can be improved effectively. Moreover, this evaluation indicator can be used as a sensitivity index to preset association gate before the loss of target or deviation of target association. Thus, not only validated measurements existing in the gate can be ensured, but also the interference from clutters and measurements can be reduced. Finally, it is verified via simulation results that the proposed optimization can effectively improve association performance and tracking accuracy.