Image-Based Tracking Control for Noncooperative Target With Visibility Maintenance

This article proposes a robust tracking control framework for noncooperative targets with visibility maintenance in GPS-denied environments. First, a variable-dimension Kernelized correlation filter (VDKCF) is designed, which integrates a translation filter and an adaptive scale filter, eliminating the reliance on artificial patterns and enhancing the adaptability of tracking noncooperative targets in real-world scenarios. Second, an image-based visual servo (IBVS) controller is designed to track the target detected by VDKCF. More specifically, we first design a nominal IBVS controller with prescribed performance in the case of unknown target motion, and then use a control barrier function (CBF) to formulate the visibility constraint and modify the nominal control input to guarantee target visibility. Additionally, we show that the VDKCF algorithm and IBVS controller enhance each other's performance. On the one hand, VDKCF empowers the controller with the ability to track noncooperative targets and provides more precise visual feedback for the controller. On the other hand, the designed controller guarantees target visibility in complex environments, thus maintaining a high peak-to-sidelobe ratio (PSR) of VDKCF, which ensures the stability for target tracking. Finally, the effectiveness of the proposed tracking system is verified by both simulation and real-world experimental results.