Optimizing Ground Vehicle Tracking Using Unmanned Aerial Vehicle and Embedded Apriltag Design

We investigate a vision-based vehicle tracking method which allows an unmanned aeriel vehicle(UAV) to track and follow a ground vehicle in coordinated Air-Ground operations. The UAV detects a customized Apriltag attached on the ground vehicle and estimates its motion status. We use a novel embedded-tag design to allow the UAV detect and lock the ground vehicle in both high and low altitude. The dynamic range achieved is from 0.3 meters to 10 meters. The motion-status estimation is improved using an optimum tag selection (as observation source) algorithm and a modified particle filter algorithm. The proposed algorithm improves the motion estimate accuracy significantly by 20%.