Collision-free manipulation of a robotic arm using the MS Windows Kinect 3D optical system

This article presents the collision-free manipulation of a robotic arm mounted to a mobile robotic unit, using the MS Windows Kinect 3D optical camera system. The 3D optical system is used to recognize objects, aiding the collision-free manipulation of the said objects with the arm of a mobile robotic system (MRS). The placement of the optical system directly on the robotic arm is essential to ensure the autonomy of the overall system. The MRS is able to recognize pre-defined objects in three dimensional space and approach and manipulate these automatically using its robotic arm. The aim of this article is to present the algorithms used in the mobile robot to guarantee the collision-free manipulation of the MRS with objects. The novelty of our approach lies in the a priori collision avoidance strategy, instead of solving collision states as they occur. In addition to this, the article presents the localization of objects in space. For this, randomly placed balls are localized by the MRS, estimating their real coordinates. Instead of using the standard RGB colorspace, we propose to utilize the HSV colorspace to assess object coordinates, since it provides more consistent results in case the ambient lightning intensity varies.