Testbed for automotive indoor localization

To enable intelligent vehicular indoor applications highly accurate localization is required. There is a high number of different approaches which are however difficult to compare due to a lack of a common evaluation methodology. Thus, we present a testbed for vehicular indoor localization, to enable the benchmarking of different approaches under comparable conditions. As a first building block, we present a simple yet highly accurate ground truth system based on off-the-shelf infrastructure cameras and printable markers. Our employed marker detection algorithm and systematic 3-layer projection approach achieves a median accuracy of 0.48cm and 0.05 degrees for 2D position and orientation.