Sensor Failure Detection and Identification (FDI) is a standard method in many technical applications. Presented paper describes the usage of such approach in autonomous mobile robot navigation tasks. The FDI method uses pattern matching principle based on Markov chain theory. Such approach is capable to avoid any miss-readings or sensor failures during navigation task. This method significantly improves the robustness of mobile robot navigation in crowded environment shared with people. Practical verification experiments were performed on autonomous mobile robot Advee, which is designed to safely move among people.