Novel signalling systems are expected to use Digital Maps to implement ATO functions or to augment technologies such as EGNSS, inertial sensors or optical sensors for localization purposes. However, the construction of a Digital Map with current technologies requires expensive railway surveying campaigns. Additionally, the cost for their maintenance and update is a major barrier to their widespread adoption. With regard to the design of Digital Maps and trackside signalling subsystems, they are based on the knowledge of two types of data: (a) the longitudinal positions of Points Of Interests with respect to the track centreline and (b) the grades or slopes of the track segments adjacent to them. Therefore, as a first step, such designs require the acquisition or the verification of the data related to POIs (i.e. nominal values of locations and gradients along with their related accuracies) and this causes extra costs and time-consuming surveying activities leading to delays in the commissioning of the lines.
With regard to the evolution of CCS systems, a new verification infrastructure is also needed to characterize new on-board solutions in terms of train position, speed and acceleration errors, as well as to assess their expected performance. The performance analysis of signalling systems is in most of previous work analysed with project dependent mechanisms and procedures. Unfortunately, there is no agreed methodology for building “reference ground truth data” in railway applications. Up to now, the investigated Ground Truth methodology has required the use of the track database and the availability of absolute and relative reference measurement systems associated with the lines for which the Ground Truth had to be built. Moreover, these methodologies also suffer from other two limitations, i.e. they provide (a) “reference ground truth data” for the train position only and (b) no information about the accuracy of the provided “reference ground truth data”.
RAILGAP developed a methodology and related toolset for building and maintaining Digital Maps based on the post-processing of data recorded by trains equipped with COTS sensors like GNSS receivers, cameras, IMUs, and LIDARs. On the other side, a similar approach and development has been used to provide high accuracy reference data for carrying out error and performance analysis. The RAILGAP project reached three main objectives. The first one is the development of an innovative Ground Truth methodology and related toolset that overcome the above described constraints and limitations. The second objective is to perform field surveys by using measured physical entities collected with trains during commercial services for building and maintaining Digital Maps. The third one is the use of AI techniques to off-line process big amount of collected data to allow the extraction of the information required for building and maintaining Digital Maps.