Vision-Based Trajectory Tracking on the 3D Virtual Space for a Quadrotor

摘要

In this paper, we propose a vision-based system for real-time 3D trajectory tracking. The vision system is implemented using the KINECT sensor (color and depth cameras). Image processing is performed using freeware and open source software. The image processing recovers the 3D position vector for the centroid of the target to track. We define a target object to track in a 3D space in order to emulate, on a virtual reality environment, a quad rotor which describes the trajectory traced on the 3D scene. A state feedback control scheme for the quad rotor is implemented on Simulink, on which information from the KINECT sensor is supplied as reference signals. The color-based object tracking approach is used. Experimental results validate our proposed method.