A robust data association for simultaneous localization and mapping in dynamic environments

JPDA (Joint Probabilistic Data Association) is vastly regarded as a more tractable and suboptimal method for solving ambiguity in data association problem in the presence of clutter for simultaneous localization and mapping (SLAM). However, JPDA generally has problems for detecting moving objects and distinguishing the new landmarks from clutter, which cause false data associations in dynamic environments. We propose a semi-temporal algorithm using three-scan JPDA to accurately correlate the observation with its corresponding landmark, and initialize the new landmark. The existence of moving clutter in validation gates is alerted by a statistic motion detector that enhances data association in a dynamic environment. This method can be applied for real-time SLAM applications with less complexity comparing with other high-cost optimal Bayesian filter. Simulation is performed to verify the effectiveness of method.