Miltos Kyriakidis

A human factors perspective on automated driving

ABSTRACT Automated driving can fundamentally change road transportation and improve quality of life. However, at present, the role of humans in automated vehicles (AVs) is not clearly established. Interviews were conducted in April and May 2015 with 12 expert researchers in the field of human factors (HFs) of automated driving to identify commonalities and distinctive perspectives regarding HF challenges in the development of AVs. The experts indicated that an AV up to SAE Level 4 should inform its driver about the AVs capabilities and operational status, and ensure safety while changing between automated and manual modes. HF research should particularly address interactions between AVs, human drivers and vulnerable road users. Additionally, driver-training programmes may have to be modified to ensure that humans are capable of using AVs. Finally, a reflection on the interviews is provided, showing discordance between the interviewees’ statements – which appear to be in line with a long history of HFs research – and the rapid development of automation technology. We expect our perspective to be instrumental for stakeholders involved in AV development and instructive to other parties.

Acceptance of Driverless Vehicles: Results from a Large Cross-National Questionnaire Study

Shuttles that operate without an onboard driver are currently being developed and tested in various projects worldwide. However, there is a paucity of knowledge on the determinants of acceptance of driverless shuttles in large cross-national samples. In the present study, we surveyed 10,000 respondents on the acceptance of driverless vehicles and sociodemographic characteristics, using a 94-item online questionnaire. After data filtering, data of 7,755 respondents from 116 countries were retained. Respondents reported that they would enjoy taking a ride in a driverless vehicle (mean = 4.90 on a scale from 1 = disagree strongly to 6 = agree strongly). We further found that the scores on the questionnaire items were most appropriately explained through a general acceptance component, which had loadings of about 0.7 for items pertaining to the usefulness of driverless vehicles and loadings between 0.5 and 0.6 for items concerning the intention to use, ease of use, pleasure, and trust in driverless vehicles, as well as knowledge of mobility-related developments. Additional components were identified as thrill seeking, wanting to be in control manually, supporting a car-free environment, and being comfortable with technology. Correlations between sociodemographic characteristics and general acceptance scores were small (<0.20), yet interpretable (e.g., people who reported difficulty with finding a parking space were more accepting towards driverless vehicles). Finally, we found that the GDP per capita of the respondents’ country was predictive of countries’ mean general acceptance score ( across 43 countries with 25 or more respondents). In conclusion, self-reported acceptance of driverless vehicles is more strongly determined by domain-specific attitudes than by sociodemographic characteristics. We recommend further research, using objective measures, into the hypothesis that national characteristics are a predictor of the acceptance of driverless vehicles.

The Deployment of Advanced Driver Assistance Systems in Europe

Advanced driver assistance systems (ADASs) are expected to significantly enhance driving safety, comfort and efficiency. This paper analyses the deployment of the ADASs in EU 28, in relation to the countries GDPs and fatalities. In addition, we explored the relation of ADASs sold with vehicle price and mass. For the analysis two sets of data were used. The first was obtained from the iMobility 2013 iCar implementation status survey, on the EU 28 ADASs deployment rates for vehicles sold in 2012. The second was derived from of a Dutch leasing company and contains information on the number and type of ADASs ordered for the Dutch and German market in 2013-2014. Results show that the deployment rate averaged over EU 28 is 2.7-12.6% for five safety related ADASs and 23% for eco driving support. At country level, results indicate that in richer countries the ADAS deployment rates are significantly higher (ρ = .758, p The leasing company data showed significant correlations between the price of vehicles and type of ADASs on board. Premium vehicles are significantly more equipped with Adaptive Cruise Control and Emergency Braking, compared to the mass market vehicles, with correlations ρ = .585 (p

Railway Accidents Caused by Human Error: Historic Analysis of UK Railways, 1945 to 2012

Accident investigation and analysis are key to reinforcing and improving railway safety. Many railway accidents have been caused by degraded human performance and human error, and the tasks of train drivers and signalers have remained essentially the same. Although new technologies and equipment have gradually reduced railway operation accidents, no investigation has been conducted to investigate whether railway performance shaping factors (R-PSFs), attributed to degraded human performance, have changed or remained constant. Focusing on UK railways, this paper analyzes railway accidents involving human error for the period 1945 to 2012. The purpose of the analysis is twofold: to identify whether the number and type of factors that affect human performance and contribute to human errors have changed during this period and to assess the quality of data collected by investigation reports and to determine whether the collection of such data has evolved and improved. The analysis identifies the number of R-PSFs and their relationship to such variables as responsible personnel, immediate causes, and the time and location of the event. The contribution of those variables to the severity of an accident is calculated. Moreover, any changes in trends and patterns in the number of R-PSFs over time are explored through smoothing techniques of time series analysis. Finally, the quality of the collected data is analyzed with the data quality index. Results show that data investigation and data collection have significantly improved. However, although accident rates have decreased, the average number and type of factors that affect human performance have remained the same.

Data Based Framework to Identify the Most Significant Performance Shaping Factors in Railway Operations

Human performance is a major contributor to railway incidents and accidents. The literature shows that it is the train drivers, signallers and controllers who mainly affect the network in terms of safety. Several studies have been conducted to investigate operators’ influence on the railway system. However, most of them are based on studies from other domains, which are not well suited and can be difficult to apply reliably to railway specific operations. On account of that, this paper presents a framework to identify the most significant performance shaping factors that influence railway operators’ performance. These factors, broadly known as performance shaping factors (PSFs), have been derived from an extensive literature review in addition to the analysis of 479 railway operational incidents and accidents that happened the last 15 years worldwide. Based on the results the existing Railway Performance Shaping Factors (R-PSFs) taxonomy is validated and its updated version is obtained. The analysis focuses on the R-PSFs and their relation to the type of events, as well as to a number of factors that affect the safety of railway operations including: the responsible personnel, the immediate causes, the time and the location of the event. Findings show that 12 R-PSFs account for more than 90% of the occurrences, regardless the severity of the event. For those R-PSFs, Pearson’s chi-square tests indicate the pair-wise associations amongst the data, while log linear analysis identifies if any higher order associations exists amongst them. Finally, regression analysis is conducted to estimate the likelihood of an event to occur due to a certain R-PSF. Results indicate the contribution of each individual R-PSF to the occurrence of a railway incident or accident. Findings will be used to direct resources more efficiently towards the development of sound solutions for improving operators performance.

The human performance railway operational index—a novel approach to assess human performance for railway operations

Abstract Human error and degraded human performance are associated with more than 80% of all railway accidents worldwide. Research on human performance and human reliability has highlighted the importance of the contextual factors associated with human errors, known as performance shaping factors (PSFs). A major shortcomings of current Human Reliability Analysis techniques, which employ qualitative and quantitative methods for assessing the human contribution to risk, lies with their little capability to model the dependencies among PSFs and to quantify their impact on human performance. This paper presents a novel approach to assess human performance accounting for the dependencies among the relevant PSFs, referred to as Human Performance Railway Operational Index (HuPeROI). The HuPeROI is developed on the integration of the Analytic Network Process and Success Likelihood Index Methodology, using the insights of 52 front-line, managerial and human factors railway personnel, and was demonstrated in three different types of railway operations: regional, high-speed and underground. Findings show that the HuPeROI can be efficiently used to assess operators’ performance as function of the quality of the relevant R-PSFs. Regulatory bodies and other stakeholders can implement the framework within their safety management systems to improve safety of railway operations.

Validity and reliability of naturalistic driving scene categorization Judgments from crowdsourcing

A common challenge with processing naturalistic driving data is that humans may need to categorize great volumes of recorded visual information. By means of the online platform CrowdFlower, we investigated the potential of crowdsourcing to categorize driving scene features (i.e., presence of other road users, straight road segments, etc.) at greater scale than a single person or a small team of researchers would be capable of. In total, 200 workers from 46 different countries participated in 1.5days. Validity and reliability were examined, both with and without embedding researcher generated control questions via the CrowdFlower mechanism known as Gold Test Questions (GTQs). By employing GTQs, we found significantly more valid (accurate) and reliable (consistent) identification of driving scene items from external workers. Specifically, at a small scale CrowdFlower Job of 48 three-second video segments, an accuracy (i.e., relative to the ratings of a confederate researcher) of 91% on items was found with GTQs compared to 78% without. A difference in bias was found, where without GTQs, external workers returned more false positives than with GTQs. At a larger scale CrowdFlower Job making exclusive use of GTQs, 12,862 three-second video segments were released for annotation. Infeasible (and self-defeating) to check the accuracy of each at this scale, a random subset of 1012 categorizations was validated and returned similar levels of accuracy (95%). In the small scale Job, where full video segments were repeated in triplicate, the percentage of unanimous agreement on the items was found significantly more consistent when using GTQs (90%) than without them (65%). Additionally, in the larger scale Job (where a single second of a video segment was overlapped by ratings of three sequentially neighboring segments), a mean unanimity of 94% was obtained with validated-as-correct ratings and 91% with non-validated ratings. Because the video segments overlapped in full for the small scale Job, and in part for the larger scale Job, it should be noted that such reliability reported here may not be directly comparable. Nonetheless, such results are both indicative of high levels of obtained rating reliability. Overall, our results provide compelling evidence for CrowdFlower, via use of GTQs, being able to yield more accurate and consistent crowdsourced categorizations of naturalistic driving scene contents than when used without such a control mechanism. Such annotations in such short periods of time present a potentially powerful resource in driving research and driving automation development.