Improved Kernel Correlation Filter Based Moving Target Tracking for Robot Grasping
Fang Peng, Qinyi Xu, Yifei Li, Maoxi Zheng, Hang Su
Abstract
Tracking and grasping moving objects is a hot topic in the field of robots, which provide great potential in the industrial scene and human-computer cooperation. Based on kernel correlation filter and vision 3D reconstruction, this paper proposes a visual-based tracking and grasping method for moving targets. An improved algorithm based on kernel correlation filter is proposed for object tracking. A scale pool is constructed and a scale filter is trained to solve the problem of algorithm scale adaptation. At the same time, the judgment mechanism of tracking results, secondary detection, and modification update mechanisms are added to improve the robustness of the algorithm. Combined with RGB-D camera, the target is reconstructed to obtain the 3D pose the target. The tracking and intercepting strategy is adopted to grasp the moving target. The proposed method is proven to have good performance through data set comparison tests and experiments on real robot systems.