Localization Integrity for Intelligent Vehicles Through Fault Detection and Position Error Characterization

Localization integrity consists in providing a real-time measure of the level trust to be placed localization estimates as vehicles operate. It provides means knowing whether position are usable for navigation purposes. This paper formalizes concept and its underlying principles. Vehicles operate different environments, so multiple sensors used ensure required performance. Different sources error exist. They must bounded according acceptable risk application. presents generic approach addressing integrity. combines measurement rejection (for measurements considered faults) characterization. For this purpose, multi-sensor data fusion with Fault Detection Exclusion algorithm is constituted using bank information filters. These filters allow detected faults isolated without any prior assumption regarding number simultaneous errors. In addition, external expressed Protection Level solution. uses Student’s $t$ -distribution order bound distribution applicable small risks after learning step. The tested on acquired public roads an experimental vehicle equipped off-the-shelf proprioceptive exteroceptive together HD map. results obtained validate proposed approach.