RoboKinect — A low-cost mobile vision system for 2.5D object detection

The new generation of 3D imaging sensors allows to go beyond the traditional 2D object detection. In this paper a low-cost robotic platform exploiting the sensing capabilities of the Microsoft Kinect is presented. The developed mobile vision system is called RoboKinect and it utilises affordable hardware components such as the Texas Instruments OMAP-based BeagleBoard and the MSP430 microcontroller, as well as open-source software. In addition we implemented a novel technique for 2.5D object detection exploiting the depth map from the Microsoft Kinect. We present also some results from the preliminary testing of the algorithm on the mobile platform.