A Hybrid Data Association Strategy in SLAM

Abstract

Data association is one of the key issues in SLAM. Against the accuracy of NN data association is low, but its efficiency is high and JCBB data association have high accuracy, but its efficiency is low, a hybrid data association strategy is proposed. At first NN produce an assumption H, according to conditions to decide whether we correct it or take what kind of corrective actions to get new assumption. Simulation results show that, both static environment and dynamic environment, this method is effective, the correlation results are reliable and its time performance is significantly better than JCBB.