Fault detection in roll angle estimation for two-wheeled vehicles

The knowledge of the lean angle represents a fundamental information to design a control system for the out-of-plane dynamics of 2-wheel vehicles. Recent papers has tackled successfully the problem of lean angle estimation by the use of inertial sensors. However, in real application, a control system needs not only the actual measure but also the information about its reliability, in order to drive a safety relevant action in the case of an improper information. This paper is focused on the study of the fault analysis of a lean angle estimation system based on the use of 4-deegre-of-freedom inertial sensor. In particular this paper adresses a fault taxonomy to analyze the fault-detection problem and presents a possible solution. Some tests on a real motorcycle, where both coherence and hardware faults are artificially generated, show the effectiveness of the aforementioned strategy.