Visual tracking and grasping of moving objects and its application to an industrial robot

摘要

In industrial robots, the use of vision sensors is indispensable, which can enhance the robot's industrial intelligence and better help the industrial robot accomplish tasks. But achieving the accuracy and time efficiency together still is a challenge in industrial tasks. In this paper, a new method is presented to track and grasp the moving object. First, the high-resolution depth and RGB sensing are acquired by Kinect v2. Next, a tracking algorithm of improved spatio-temporal context is applied to tracking the moving object, and the gripper position in the base coordinate system is calculated. To accomplish grasping tasks, a linear prediction method is applied to predict the trajectory of the moving object in three dimension space, and the distance between the moving object and the gripper are constantly decreased by a simple grasping strategy. Finally, the tracking system based on the industrial robot is set up in our laboratory. The effectiveness of the proposed method is verified.