The tracking of objects for an autonomous vehicle requires sufficiently reliable data processing and association. In this paper, the signal data processing of sensed LIDAR and the multiple target track management algorithm of a maneuvering vehicle are presented. The algorithm is employed for 2D LIDAR sensor mounted in a moving vehicle and navigating in a high-way. The adaptive segmentation, feature creation from point cloud, data association and prediction modelling are the key features of track management. Initialization of the track has been developed based on constant velocity model hypothesis in order to facilitate target management in a high-way crowded environment. The multiple target tracking are associated with feature identification of the targets and also prediction modelling of moving occluded object. The prediction model of moving vehicles and pedestrians are the focus area of this research. The non-parametric probabilistic data association approach has been introduced in the paper for modelling of track management and update of state covariance for moving targets with different characterized features and velocities. The probabilistic data association is applied on the classified feature point of the moving targets. Track management algorithm distinguishes false positive target signals and clutters in the environment using object classification methodology and thereby reduces the probability of track loss due to sensor noise. The adaptive segmentation with object classification and probabilistic data association approach results in an efficient track management system. The position correction of tracking object due to orientation change of the host vehicle has been addressed as an additional feature to this research. The inertial measurement unit based orientation compensation algorithm is set to enhance the track management of objects even in a cluttered environment.