Improving low-cost GNSS navigation in urban areas by integrating a Kinect device

In the last decades, low-cost GNSS receivers have been widely used for navigation purposes. Some of them deliver also raw data, allowing for a more sophisticated processing, such as the double-difference approach, and therefore a more accurate positioning, typically at the decimeter level. However, these accuracies can be generally achieved only with a good sky visibility, that is a critical issue in urban areas even using low-cost receivers equipped with a high-sensitive antenna. In this respect, a significant contribution comes from the use of digital images or dense point clouds which provides an estimate of the sensor kinematic position. To maintain the low-cost target, the Kinect device, endowed with RGB and depth cameras, can be used. In this work, we have first processed the GNSS raw data from a u-blox receiver by using the free and open source goGPS software. Then, we have studied the integration of the Kinect device by a proper Kalman filter. An outdoor experiment has been arranged with the aim of testing the hardware and software system.