New Approach to Multitarget Tracking Using Probabilistic Data Association

摘要

This report develops the theory for a multitarget tracking algorithm based on Probabilistic Data Association with new selection rules for assigning sensor measurements to target track and for forming multitrack clusters. These new rules remove the requirement to form a gate about each target's predicted position for the selection of sensor measurements. The resultant algorithm is the same for all target tracks and clutter conditions. The algorithm adapts to the sensor measurements via probability terms which model the environment and sensor processing.