Quantifying drivers’ perceived risk is important in the design and evaluation of the behaviour of automated vehicles (AVs) and in predicting takeovers by the driver. A ‘Driver's Risk Field’ (DRF) function has been previously shown to be able to predict manual driving behaviour in several simulated scenarios. In this paper, we tested if the DRF-based risk estimate (rˆ) could predict manual driving behaviour and the driver's perceived risk during automated driving. To ensure that the participants perceived realistic levels of risk, the experiment was conducted in a test vehicle. Eight participants drove five laps manually and experienced 12 different laps of automated driving on a test track. The test track consisted of three sections (which were sub-divided into 12 sectors): curve driving (9 sectors), parked car (1 sector), and 90-degree intersections (2 sectors). If the driver verbally expressed risk or performed a takeover, that particular sector was labelled as risky. The results show that the DRF risk estimate (rˆ) predicted manual driving behaviour (ρ_steering=0.69, ρ_speed=0.64), as well as correlated with the driver's perceived risk in curve driving (r² = 0.98) and while negotiating a car parked outside the lane boundary (r²=0.59). In conclusion, the DRF-based risk estimate (rˆ) is predictive of manual driving behaviour and perceived risk in automated driving. Future research should include tactical and strategic components to the driving task.

@en

The National Highway Transportation Safety Administration (NHTSA) reports that 94-96% of the road accidents involve human error. These statistics make it seem as if humans are terrible drivers. However, a different set of numbers paint a completely different picture. According to the United States Bureau of Transportation Statistics, the failure rate of human drivers is 0.68 fatalities per 100 million kilometres. This number is so low that autonomous vehicle manufacturers are having a hard time proving that their vehicles are safer than human drivers. The safety benefits of Driver Assistance Systems, autonomous vehicles, and other safety systems are not being challenged here. However, it needs to be realised that a non-fatigued, attentive human is one of the safest drivers we can have....@en

Current driving behaviour models are designed for specific scenarios, such as curve driving, obstacle avoidance, car-following, or overtaking. However, humans can drive in diverse scenarios. Can we find an underlying principle from which driving behaviour in different scenarios emerges? We propose the Driver’s Risk Field (DRF), a two-dimensional field that represents the driver’s belief about the probability of an event occurring. The DRF, when multiplied with the consequence of the event, provides an estimate of the driver’s perceived risk. Through human-in-the-loop and computer simulations, we show that human-like driving behaviour emerges when the DRF is coupled to a controller that maintains the perceived risk below a threshold-level. The DRF model predictions concur with driving behaviour reported in literature for seven different scenarios (curve radii, lane widths, obstacle avoidance, roadside furniture, car-following, overtaking, oncoming traffic). We conclude that our generalizable DRF model is scientifically satisfying and has applications in automated vehicles.

@en

Gibson and Crooks (1938) argued that a ‘field of safe travel’ could qualitatively explain drivers' steering behavior on straights, curved roads, and while avoiding obstacles. This study aims to quantitatively explain driver behavior while avoiding obstacles on a straight road, and quantify the ‘Driver's Risk Field’ (DRF). In a fixed-based driving simulator, 77 (7 longitudinal and 11 lateral) positions of the obstacles were used to quantify the subjectively perceived and objectively (maximum absolute steering angle) measured DRF for eight participants. The subjective response was a numerical answer to the question “How much steering do you think you need at this moment in time?” The results show that the propagation of the width of the DRF, along the longitudinal distance, resembled an hourglass shape, and all participants responded to obstacles that were placed beyond the width of the car. This implies that the Driver's Risk Field is wider than the car width.

@en

When drivers encounter a road narrowing two potential adaptation strategies come into play that may increase safety margins: decreasing speed and increasing neuromuscular stiffness of the arms. These two adaption strategies have so far been studied in isolation. We expect that there is a trade-off between these two strategies, and that risk duration would impact a driver's selection of the trade-off. Specifically, we hypothesized that for a short risk duration, drivers will favour increased neuromuscular stiffness over speed reduction; and vice versa for longer risk durations. Twenty-six participants drove in a driving simulator and encountered different risk durations; realized by road narrowings (from 3.6 m to 2.2 m) of varying lengths (10 m, 100 m, 250 m, and 500 m). The neuromuscular stiffness was quantified by measuring the grip force exerted by both hands. The results show that all road narrowing conditions successfully induced driver adaptations, as a significant reduction in speed and increase in grip force was observed. However, the tested drivers did not consistently select the hypothesized different trade-offs for increasing duration of road narrowing: a low correlation was found between speed and grip force adaptations. Interestingly, individual trade-off were consistent: the within-subject variability in speed-grip force adaptations was low across the tested risk durations. Future research should further elucidate the underlying motivations for these individual adaptation strategies.

@en

Current haptic control systems provide feedback torques based on a lateral deviation with respect to a reference trajectory (i.e., centre of the lane), which do not capture the satisficing behaviour human beings typically adopt during a lane keeping task. As such, a novel time-to-lane-crossing-based controller is proposed, which is expected to provide more human-like guidance. The aim of this study is to describe a novel time-to-lane-crossing-based controller and investigate its potential as an alternative to previous reference-trajectory-based guidance. In a simulator study twenty-four participants drove three trials through a single-lane, 10.8 km long road (width: 3 m), receiving three types of guidance, namely 1) none (manual), 2) reference based controller, 3) TLC-based controller. Results showed that both the reference-based, as well as the TLC-based guidance provided significant safety benefits, in terms of more centred and less varying lane position, and higher safety margins. Moreover, no significant differences were revealed between the two guidance approaches. In conclusion, the TLC-based guidance is a potential alternative to reference trajectory-based guidance. Nevertheless, a more detailed analysis is warranted to investigate the two approaches in different driving conditions, like road width, straights, and curves.@en

The purpose of this study is to develop and validate a human-like steering model that can capture, not only the mean, but also the intradriver variability (IDV) of steering behavior, in both routine and emergency scenarios. The IDV model proposed in this study is based on the assumption that steering behavior, in both scenarios, is governed by the same principles as performing point-to-point reaching tasks. The optimal feedback control framework that models the reaching tasks, and the presence of signal-dependent noise in motor commands and sensory feedback, are the mainstays of the proposed model. The driver is assumed to have acquired an internal model of system (muscles, arms, and vehicle) dynamics, and has a preview of the upcoming road. The model is validated using simulator-based data from both routine (curve negotiation) and emergency (obstacle avoidance) scenarios. The IDV model could capture mean steering torque behavior in both routine (variance accounted for (VAF) <formula><tex>$=$</tex></formula> 92<formula><tex>$\%$</tex></formula>) and emergency (VAF <formula><tex>$=$</tex></formula> 74<formula><tex>$\%$</tex></formula>) scenarios, but more prominently, it could capture the standard deviation of the steering torque as well, in both routine (VAF <formula><tex>$=$</tex></formula> 83<formula><tex>$\%$</tex></formula>) and emergency (VAF <formula><tex>$=$</tex></formula> 65<formula><tex>$\%$</tex></formula>) scenarios. The promising results show that including signal-dependent noise and modeling steering as a reaching task are steps in the right direction in the field of driver modeling. The model, however, poorly captured the lateral deviation behavior, primarily suspected due to the satisficing behavior exhibited by humans. Developing a nonlinear-iterative version of the IDV model could address the limitations.

@en

The interaction between a human driver and an automated driving system may improve when the automation is designed in such a way that it behaves in a human-like manner. This paper introduces a human-like steering model, in which the driver adapts to the risk due to uncertainty in the environment. Current steering models take a risk-neutral approach, while the fields of economics and sensorimotor control suggest that humans exhibit risk-sensitive behavior. The proposed model uses a risk-sensitive optimal feedback control structure to predict steering behavior. The paper studies the effect of the risksensitivity parameter and compares the prediction of the riskneutral and risk-sensitive controllers in a simulated abstraction of two scenarios: (a) driving while being subjected to lateral wind gusts and (b) overtaking an unpredictably swerving car. The simulation results show that the risk-sensitive model adapts to the uncertainty in the environment. Experimental data will be needed to validate the predictions of our model.

@en