Tracking uncertain moving objects using dynamic track management in Multiple Hypothesis Tracking

Laser range finder (LRF) has been widely used for detecting and tracking moving objects. In autonomous navigation, LRF provides reliable data of moving objects surrounding the vehicle for obstacle avoidance. Data association is a crucial process for a successful moving objects tracking. In urban area, objects tend to move in various directions, thus increasing the possibility of incorrect data associations. In this paper, a reliable dynamic track management (DTM) based on Multiple Hypothesis Tracking (MHT) method is proposed. The Interacting Multiple Model (IMM) with Kalman filter provides extra information for track management process which increases the performance of data association. Simulations and real time experiment were conducted to evaluate the proposed track management in various scenarios to deal with the creation of new track, track deletion and detection of cross track. The results suggested that the proposed method produced acceptable results, reflecting the accuracy of object identification for all moving objects in all tested scenarios.