ASCE-ASME Journal of Risk and Uncertainty in Engineering Systems, Part B: Mechanical Engineering

This article is available in the ASME Digital Collection at .

Safety analysis is a significant step for safety-critical system development. Compared with traditional vehicles, the system interactions for autonomous vehicles are more abundant and complex. Traditional hazard analysis methods, such as failure mode and effects analysis (FMEA) and fault tree analysis (FTA) which are based on the component failure and reliability theory, cannot identify the system hazards related to system interactions. An emerging hazard analysis method based on systems theory, systems theory process analysis (STPA) mainly focuses on identifying the control system hazards caused by system interactions. In this study, STPA method is used to identify the potential hazards and casual factors for autonomous emergency braking systems by concentrating on system interactions. To improve the consistency between system design and safety analysis, the workflow combining model-based systems engineering (MBSE) and STPA is proposed. The systems modeling language (SysML) is used to describe control structure and system interaction relationships. According to the identified casual factors, certain constraints and requirements can be derived, which can provide the guidance for system development with respect to system design. Furthermore, the quantitative analysis of certain unsafe control action is conducted by simulation, which shows effectiveness and feasibility of the proposed method in safety analysis and system design. This article is available in the ASME Digital Collection at .

Modern technical systems consist of heterogeneous components, including mechanical parts, hardware, and the extensive software part that allows the autonomous system operation. The heterogeneity and autonomy require appropriate models that can describe the mutual interaction of the components. uml and sysml are widely accepted candidates for system modeling and model-based analysis in early design phases, including the analysis of reliability properties. uml and sysml models are semiformal. Thus, transformation methods to formal models are required. Recently, we introduced a stochastic dual-graph error propagation model (DEPM). This model captures the control and data flow structures of a system and allows the computation of advanced risk metrics using probabilistic model checking techniques. This article presents a new automated transformation method of an annotated state machine diagram (SMD), extended with activity diagrams (ADs), to a hierarchical DEPM. This method will help reliability engineers to keep error propagation models up to date and ensure their consistency with the available system models. The capabilities and limitations of the transformation algorithm are described in detail and demonstrated on a complete model-based error propagation analysis of an autonomous medical patient table (MPT). This article is available in the ASME Digital Collection at .

Light detection and ranging (lidar) imaging systems are being increasingly used in autonomous vehicles. However, the final technology implementation is still undetermined as major automotive manufacturers are only starting to select providers for data collection units that can be introduced in commercial vehicles. Currently, testing for autonomous vehicles is mostly performed in sunny environments. Experiments conducted in good weather cannot provide information regarding performance quality under extreme conditions such as fog, rain, and snow. Under extreme conditions, many instances of false detection may arise because of the backscattered intensity, thereby reducing the reliability of the sensor. In this work, lidar sensors were tested in adverse weather to understand how extreme weather affects data collection. Testing setup and algorithms were developed for this purpose. The results are expected to provide technological validation for the commercial use of lidar in automated vehicles. The effective ranges of two popular lidar sensors were estimated under adverse weather conditions, namely, fog, rain, and snow. Results showed that fog severely affected lidar performance, and rain too had some effect on the performance. Meanwhile, snow did not affect lidar performance. This article is available in the ASME Digital Collection at .

As autonomous vehicle (AV) intelligence for controllability continues to develop, involving increasingly complex and interconnected systems, the maturity level of AV technology increasingly depends on the systems reliability level, also considering the interactions among them. Hazard analysis is typically used to identify potential system risks and avoid loss of AV system functionality. Conventional hazard analysis methods are commonly used for traditional standalone systems. New hazard analysis methods have been developed that may be more suitable for AV system-of-systems complexity. However, a comprehensive comparison of hazard analysis methods for AV systems is lacking. In this study, the traditional hazard analysis methods, hazard and operability (HAZOP) and failure mode and effects analysis (FMEA), as well as the most recent methods, like functional resonance analysis method (FRAM) and system-theoretic process analysis (STPA), are considered for implementation in the automatic emergency braking system. This system is designed to avoid collisions by utilizing the surrounding sensors to detect objects on the road, warning drivers with alerts about any collision risk, and actuating automatic partial/full braking through calculated adaptive braking deceleration. The objective of this work is to evaluate the methods with the unified theory of acceptance and use of technology (UTAUT) approach, in terms of their applicability to AV technologies. The advantages of HAZOP, FMEA, FRAM, and STPA, as well as the possibility of combining them to achieve systematic risk identification in practice, are discussed. This article is available in the ASME Digital Collection at .

Modern vehicles are connected to the network and between each other through smart sensors and smart objects commonly present on board. This situation has allowed manufacturers to send over-the-air updates, receive diagnostic information, and offer various multimedia services. More generally, at present, all this is indicated by the term “Vehicle to Everything” (V2X), which indicates a system of communication between a vehicle and any entity that may influence the vehicle and vice versa. However, it introduces problems regarding the vehicle's IT security. It is possible, for example, by tampering with one of the electronic control units (ECUs) to take partial or total control of the vehicle. In this paper, we introduce a preliminary study case of a probabilistic approach in an intrusion detection system over the CAN-bus to guarantee cybersecurity inside connected vehicles. In particular, through the use of an innovative two-step detection algorithm that exploits both the variation of the status parameters of the various ECUs over time and the Bayesian networks can identify a possible attack. Starting from a domain analysis is possible to find out what are the parameters of interests and how these are related to each other. The first experimental results seem encouraging. This article is available in the ASME Digital Collection at .

Dynamic window approach (DWA) is one of the most widely used algorithms for local path planning and autonomous navigation. Although many successful examples have been shown under various operation conditions, to the authors' best knowledge, there is a lack of systematic reliability analysis, its further design improvement, and systems operation guidelines for meeting reliability requirement under different operation conditions. Several goals can be defined for a successful path planning and autonomous navigation. Among them, assurance of the collision avoidance and reaching the goal with less time are pivotal requirements, yet such reliability analysis is rarely conducted in a rigorous manner. Furthermore, design improvement and systems operation design based on rigorous reliability analysis can hardly be found in this area. This paper addresses such a research gap for autonomous navigation reliability analysis and further conducts design improvement and characterizes systems operation conditions for meeting the collision avoidance reliability using the DWA. To address the technical challenges associated with limited number of simulations or experiments, reliability analysis is conducted using Bayesian statistics combined with the Monte Carlo simulation (MCS). Design improvement and reliable operation conditions can then be conducted based on the reliability analysis. Results indicate that performance reliability of the DWA is sensitive to its parameter configuration, which can be improved through reliability-based design optimization. With characterized collision avoidance reliability constraints, performance reliability of the DWA can be ensured through adjusting its operation parameters. This article is available in the ASME Digital Collection at .

A new method for reliable fatigue life prediction in metal structural components is developed, which quantifies uncertainties using interval variables. Using this crack-initiation-based method, first, the uncertainties in laboratory test data for the fatigue failure of a structural detail are enumerated. This uncertainty quantification is performed through an interval-based enveloping procedure that relates the interval stress ranges to the number of cycles to failure. This will lead to the construction of an interval S–N relationship. Next, the uncertainties in field test data are enumerated in the extremum values of each stress range, as intervals, leading to the construction of interval stress ranges. For both the laboratory and field data uncertainty analyses, the mean stress effects are considered. Next, the interval damage accumulated over the duration of the field data is determined using the constructed interval S–N relationship and the obtained interval stress ranges. Then, the interval existing damage and interval remaining life are determined. Finally, as a conservative measure, the minimum remaining fatigue life is obtained in which all uncertainties are considered. A numerical example illustrating the developed method is presented, and the results are compared with results obtained by both Monte Carlo simulation and optimization. Using this method, for the numerical example considered, it is shown that the results for bounds on the existing damage and the remaining fatigue life are sharp. Moreover, due to its set-based approach, the method is significantly more computationally efficient when compared with iterative procedures. This article is available in the ASME Digital Collection at .

A novel method is proposed in this work for the classification of fault in the ball bearings. Applications of K-nearest neighbor (KNN) techniques are increasing, which redefines the state-of-the-art technology for defect diagnosis and classification. Vibration characteristics of deep groove ball bearing with different defects are studied in this paper. Experimentation is conducted at different loads and speeds with artificially created defects, and vibration data are processed using kurtosis to find frequency band of interest and amplitude demodulation (Envelope spectrum analysis). Bearing fault amplitudes are extracted from the filtered signal spectrum at bearing characteristic frequency. The decision of fault classification is made using a KNN machine learning classifier by training feature data. The training features are created using characteristics amplitude at different fault and bearing conditions. The results showed that the KNN's accuracies are 100% and 97.3% when applied to two different experimental databases. The quantitative results of the KNN classifier are applied as the guidance for investigating the type of defects of bearing. The KNN Classifier method proved to be an effective method to quantify defects and significantly improve classification efficiency. This article is available in the ASME Digital Collection at .

Corrosion in hull structure of Coast Guard cutters is a primary degradation mode that accounts for a significant portion of depot budgets and the occasional unavailability of ships in general. Corrosion exhibits great variability spatially and temporally. This paper presents, summarizes, and analyzes a one-of-a-kind dataset for end-of-life corrosion estimation and profile of ship hull structure. The dataset was created over several years and on several vessels, and collected by maintenance personnel at several geographic locations. This study analyzes wastage data due to corrosion that were systematically collected in 2007–2008 from twelve 210-foot medium endurance cutters, commissioned in 1964–1969, in the form of thickness measurement using visual inspection and ultrasonic testing methods. A total of 76,091 thickness measurements were analyzed at positions covering the entire hulls. The measured corrosion levels mean is about 0.02–0.04 in. (1 in. = 25.4 mm), i.e., 6–14% of the as-built thicknesses after no more than 43 years of use of these 12 cutters as of 2007; however, the analysis of outliers indicates that the average wastage values can be misleading in predicting extreme corrosion. A method is proposed for estimating the counts and intensity of outliers. Examining geographic locations of the operations of these cutters and corrosion revealed that southern warm water led to appreciably larger corrosion compared to the northern colder waters, at a ratio of about 1.25–1.5. This article is available in the ASME Digital Collection at .

The objective of this work was to better understand pedestrians' understanding, trust, comfort, and acceptance of autonomous vehicle (AV) external human-machine interfaces (eHMIs). A link between mechanical engineering (i.e., automotive engineering) and civil engineering (i.e., multimodal transportation systems) is necessary to understand the effectiveness of varying AV-to-human communication strategies. Using a within-subject experiment design, 47 participants interacted with AVs possessing one of four eHMIs in a virtual reality (VR) environment. We administered a Likert scale survey to measure participants' perceptions of the eHMIs and used ordinal logistic regressions to analyze the results. We also accounted for participants' gender and stated interest in AVs, novel contributions to this field of research. The presence of an eHMI was found to improve participants' perceptions of AVs. Although females generally reported higher levels of understanding, trust, comfort, and acceptance, males' scores increased more significantly with the introduction of an eHMI. Text eHMIs outperformed nontextual interfaces, with participants noting the best perceptions with the text eHMI located on the AV's grille. Participants' understanding and identification of right-of-way (ROW) were most improved by the eHMIs while trust and comfort were most impacted by the participants' stated interest in AVs. Acceptance had little response to the eHMIs or stated AV interest and gender had little impact in the statistical models. This research supports the development of a standard, uniform AV-pedestrian communication strategy and strengthens the connection between humans and AVs. This article is available in the ASME Digital Collection at .

Six years (2015–2020) of autonomous vehicle (AV Level 3) crash data from California's (CA) OL 316 collision reports of AV crashes while in the autonomous mode (AM) or disengaged from AM just before the collision, divided by the associated CA AV make and mileage driven in the AM, are compared with the University of Michigan Transportation Research Institute (UMTRI) corrected human driver crash frequency. AV test drivers in CA mandatorily self-report every crash, whereas average drivers underreport minor accidents, so the UMTRI reporting correction factor permits comparison. CA's AV AM mileage is only a few million miles over the last few years, with virtually no police-reported crash data yet available. OL 316 crash consequence data (e.g., damage, injuries, etc.) is anecdotal and inconsistently self-reported. The CA collision report data indicate the CA AV test fleet exhibits multiples of the human crash frequency. Invariably, the AV accidents are the human driver's fault, with a majority being rear collisions. The human drivers appear less able to anticipate the AV's more conservative driving. CA's AV experience predicts more widespread deployment of existing AV technologies is not likely to reduce vehicle crash frequency, at least in the short term, and might well increase it. This article is available in the ASME Digital Collection at .