Motion control of a mobile robot using kinect sensor

This article presents an algorithm for the segmentation and tracking of the hand that work in real time and have a low computational cost, so they do not need a highperformance computer to operate. The algorithms work only with images of depth (Kinect) so they are invariant to the light conditions. In the tests performed for the hand position tracking algorithm, an error of an average of 5 mm was obtained. To validate the operation a mobile robot of differential architecture was implemented in order to control the speed and angular position; The dynamic and kinematic model of the prototype was obtained to design the controllers. The functional tests were done on a computer with an Intel Xeon X5550 4-core 2.67 GHz processor, 4 GB of RAM and an NVIDIA GeForce GTX 560 Ti video card, the orientation orders depend on the position of the user's hand and they are sent to the robot controller via Bluetooth.