Fast Joint Data Association Algorithm for SLAM of Intelligent Vehicle

Data association is a difficult problem for simultaneous localization and mapping (SLAM) of intelligent vehicle. In order to obtain data association results quickly and accurately, a new fast joint data association (FJDA) algorithm was proposed in this paper. The advantages of the sequential compatibility nearest neighbor (SCNN) algorithm, which is easy to implement, and the concept of optimality of joint compatibility branch and bound (JCBB) algorithm were combined. Firstly, SCNN algorithm was used to process all measurement-feature pairs in the local map and the association results were obtained. Secondly, the accuracy of the association result was judged. If the association failed, DBSCAN algorithm was applied to divide the current measurement into several groups, and then JCBB algorithm was performed in each group. Eventually, the associated solution of each group was fused to get the final association results. The performance of the proposed algorithm, SCNN algorithm and JCBB algorithm were compared through simulation experiments. The simulation results show that the proposed algorithm has high real-time ability and high accuracy.