This project aims at exploiting the Augmented Reality (AR) technology potential in vehicles to overlay crucial information for the driver within his/her line of sight, allowing him/her to stay in the “loop” of driving keeping the eyes on the road. Nowadays, AR systems utilize Head-Mounted Displays (HMDs) to show virtual features just in static environments like rooms or workstations. These situations are the ideal circumstances for the application of localization algorithms, typically built-in HMDs. In literature, there are not solutions meant for their application in a constantly moving environment, specifically the road, seen by the driver in front of the vehicle while driving. The proposed research project aims at designing a localization algorithm for the presented use-case, integrating it in a software package. The software package should be directly deployed on an HMD and consists of two main applications. The former displays to the driver a virtual instrumental panel fixed in the cockpit, in order to show the vehicle’s information to the user while maintaining his/her gaze on the road. The latter highlights obstacles in front of the vehicle detected by an external LiDAR sensor to promptly alert the driver of an incoming danger. The developed localization algorithm exploits computer vision techniques to track fiducial features fixed inside the cockpit in a robust, reliable and efficient way. This is done through the study of images captured by the cameras integrated in the HMDs. The software package has been implemented for two different HMDs (HoloLens 1 and ThirdEye X2), and an accurate analysis has been carried on to identify the advantages and disadvantages of each device.
