Determination of Object Location for Robotic Grasping Using Depth Vision Sensor

Abstract

The object grasping is one of the basic tasks of the mobile robotic platforms equipped with robotic manipulators. This task can be divided in two smaller sub tasks. First sub task is the determination of the optimal position for the mobile robotic platform for successful execution of the second sub task. The second sub task is the actual object grasping performed by the already positioned robotic manipulator. This paper investigates how the available depth and color vision sensor can be incorporated within the robotic control in order to improve the first sub task. An approach for determination of the best suitable position of the mobile platform for object grasping. This approach combines the data gathered by the depth sensor and uses computer vision algorithms to find the correct projection of the object of interest on the two-dimensional map generated by the navigation system. It takes into consideration the parameters of both the mobile platform and the robotic manipulator. The conducted experiments on a real robotic system are discussed and these experiments demonstrate that the proposed approach improves the abilities for object grasping of the robotic system.