RobotEye Technology for Thermal Target Tracking Using Predictive Control

Thermal cameras are widely used in the fatigue analysis of mechanical structures using the thermoelastic effect. Nevertheless, such analysis is hampered due to blurry images resulting from the motion of structure-under-test. To address the issue this paper presents a system that utilizes robotic vision and predictive control. The system comprises of a thermal camera, a vision camera, a RobotEye, and a fiducial detection system. A marker is attached to a thermal target in order to estimate its position and orientation using the proposed detection system. To predict the future position of the thermal moving object, a Kalman filter is used. Finally, the Model Predictive Control (MPC) approach is applied to generate commands for the robot to follow the target. Results of the tracking by MPC are included in this paper along with the performance evaluation of the whole system. The evaluation clearly shows the improvement in the tracking performance of the development for thermal structural analysis.