A cost-effective solution for HD Maps creation

In the following paper, the authors propose an effective and reliable solution to the HD mapping problem. The solution is composed of relatively low-cost hardware components, if compared to the most advanced HD mapping solutions available on the market. The core of the solution is a robust post-processing algorithm that processes the point-cloud generated by a LiDAR, the frames collected by three cameras, the data coming from a GNSS receiver and an IMU, and the odometry data available on the car's CAN BUS. After the generation of the georeferenced point-cloud, the post processing algorithm extracts road geometries, objects and features to build HD maps. Such solution was developed and tested on our research car, and tested in Garching, near Munich, Germany. The results of the 3D mapping system are presented later, showing how the combination of low-cost technologies with a strong data fusion algorithm can provide exceptional performance.