GNSS/IMU Tightly Coupled Scheme with Weighting and FDE for Rail Applications

ION ITM 2020, International Technical Meeting of Institute of Navigation, San Diego, ETATS-UNIS, 21-/01/2020 - 24/01/2020; GNSS used as a standalone positioning system fulfils most of the requirements for many applications since decades. However, the need of high available, accurate and fail-safe positioning systems for new applications such as rail autonomous vehicles for train signaling applications motivates the community to explore novel solutions. In this context, multi-constellation multi-frequency GNSS receivers have the potential to enhance the positioning solutions. However, when applied to safety-critical land transportation system, the localization function not only needs to meet accuracy requirements, but more importantly, it also needs to bound the positioning errors in order to reduce the risk of unexpected and undetected faults. Thus, in order to reach continuous and fail-safe positioning solutions, the use of complementary heterogeneous sensors with a smart hybridization becomes essential. For safety-related and regulated applications as in railways, one will also need, as a next step, to develop the capacity of assessing the safety of the hybrid solution. The concept of integrity can help to reach this requirement and is currently investigated for rail applications. RAIM (Receiver Autonomous Integrity Monitoring) was first defined for aviation applications and is widely used for their safety assessment. However, due to the different safety requirements and operational environment in aviation and land applications (i.e. railway and road application), dedicated algorithms need to be developed for such applications. Indeed, terrestrial applications can face particularly harsh environments (urban canyon, forests ...) in which a dilemma appears between the (reduced) availability of visible satellites and the possibility of encountering multiple simultaneous errors (NLOS, multipath interferences ...). Under these conditions, implementing a strategy based solely on a FDE layer can lead to a decrease in the availability of the localization function. Conversely, relying solely on a strategy of weighting observation error models cannot constitute an acceptable solution from a safety point of view. Only a harmonious combination of these two strategies in a stringent environmental condition can achieve the availability and security requirements. With a real dataset collected along a railway line, we compare the positioning accuracy of a GNSS/IMU tightly coupled system before and after weighting and FDE schemes. The results show that the average value of Horizontal Position Error (HPE) has been reduced after applying the performance improvement method on the GNSS/IMU system. It also shows that the implemented Horizontal Protection Level (HPL) correctly bounds HPE.