Design of monitoring concepts for motion control of autonomous heavy vehicles

Abstract

Currently, the major trend in the automotive industry is to develop autonomous vehicles. All OEMs are focusing on increasing automation in their vehicles. Increase in automation leads to the design of complex safety-critical control algorithm. As the motion control controls the driving dynamics of the whole autonomous vehicle, any faults by the motion control may lead to hazardous events. So, in order to ensure safety, the outputs of the motion control should be monitored to check if the vehicle is following the intended path. The main objective of the thesis is to design an algorithm to detect faults affecting the motion on complete vehicle level. The first part of this report contains findings of a literature review on functional safety, monitoring concepts and fault detection methods. After completing the literature review on different fault detection method, it is found that monitor based on forward dynamics is best suited for this application. The monitor is modelled based on the single-track model of vehicle. It is discovered that the acceleration is more sensitive to torque faults than longitudinal velocity for monitoring longitudinal dynamics. For lateral dynamics, the yaw rate is chosen to monitor as it is found to be sensitive enough to detect faults. The designed monitor is improved by making it adaptive monitor in order to ensure safety and robustness. Suitable threshold values are composed based on the safety goals provided in order to classify as a fault. The designed monitor is validated in the simulation environment by injecting appropriate faults. From the results it is found, the monitor based on forward dynamics detects the faults in longitudinal acceleration and yaw rate quickly, thus ensures safety.