VINCI Autoroutes Foundation

REPORT CI2N Influence of a speed limiter and regulator on driving habits. Study conducted with the support of the VINCI Autoroutes Foundation for responsible driving André Dufour & Olivier Després 01/09/2013 Centre d'investigations Neurocognitives et Neurophysiologiques

2/22

TABLE OF CONTENTS 1. SUMMARY... 4 2. AIM... 6 3. METHODS... 6 4. RESULTS... 8 4.1. ANALYSES OF DRIVING ON A SIMULATOR... 8 4.1.1. Driving data related to road behavior... 8 4.1.1.1. Average speed... 8 4.1.1.2. Overtaking and pulling back... 10 4.1.2. Driving parameters sensitive to hypovigilance... 12 4.1.2.1. Lateral position and speed adjustments... 12 4.1.2.2. Deceleration distance at the occurrence of a scenario... 14 4.1.3. Summary of analyses of driving data... 17 4.2. ANALYSIS OF THE PHYSIOLOGICAL INDICATORS OF VIGILANCE... 18 4.2.1. Self-evaluation of the level of vigilance... 18 4.2.2. Alpha oscillations measured by electroencephalography... 19 4.2.3. Ocular movements... 21 5. CONCLUSION... 22 3/22

1. Summary To date, no study has assessed the ability of drivers to use non-adaptive speed regulators/limiters or the effect of these systems on driver alertness. The reduction in cognitive load provided by these systems in non-critical situations or during monotonous driving would likely result in a decrease in vigilance. Indeed, any voluntary action to control the vehicle, including speed control, would likely reactivate driver alertness. Consequently, automated speed regulation technologies that reduce the frequency of actions performed by a driver would be expected to reduce his/her vigilance and wakefulness. We hypothesized that speed regulation systems would have two main effects. First, they might represent an additional danger in so-called critical situations (e.g., increased traffic density, and situations requiring an immediate control of speed). Second, the absence of motor actions necessary to regulate speed (e.g., foot on the accelerator), and the lack of reactivating actions, can promote entry into a hypovigilant state. Ninety subjects, divided into three age groups (i.e., 18-30, 40-50, and >60 years), were tested on a driving simulator. The task consisted of a 120-kilometer ride on a motorway and included four situations inducing drivers to adjust their speeds. Each participant completed the driving task during two experimental sessions, in which he/she had to use either a speed regulator or a speed limiter. During a third (control) session, no speed regulating system was available. Analysis showed that the use of the speed regulator had significant effects on both driver vigilance and driving behavior. Indeed, physiological as well as driving parameters showed that drowsiness while driving was increased when participants used the speed regulator compared with the control condition. Self-assessment of the levels of fatigue and wakefulness also showed effects of the speed regulator, with more pronounced increases in fatigue and drowsiness after 30 minutes of driving with than without the speed regulator. In addition, use of the speed regulator increased drowsiness. After one hour of driving, drivers became slower to react at the approach of a risk event when using a speed regulator than when using a speed limiter. Interestingly, we found that drivers in the 40-50 and >60 years age groups were less affected by fatigue and by the use of a speed regulator than younger drivers. The results of these subjective self-assessments were confirmed by those obtained from the analyses of physiological indicators, such as alpha oscillations on electroencephalograms (EEG) and eye movements. 4/22

Driving behaviors were also globally affected by the use of a speed regulator. Indeed, drivers tended to shorten the distance to the vehicle ahead of them when using the speed regulator, both in freeflowing traffic and in areas of high vehicle density, especially when approaching risk zones. The speed limiter had much less effect on alertness and driving than the speed regulator. The level of vigilance, as measured by subjective self-assessment and alpha oscillations, using the speed limiter did not differ significantly from that observed under the control condition. The reduced frequency of vertical eye movements observed when using the speed limiter may simply reflect a reduction in checking the speed on the dashboard. Driving behaviors were nevertheless modified similarly to the changes observed for the speed regulator, specifically by reductions in inter-vehicle distances and longer braking latencies. 5/22

2. Aim The aim of this project was to evaluate, in three consecutive sessions of simulated driving on a motorway, the impact of the use of a non-adaptative speed regulator and a speed limiter on driving performance and vigilance. 3. Methods Ninety subjects, divided into three age groups, 30 each aged 18-30, 40-50, and >60 years, with each group consisting of 15 women and 15 men, were tested on a driving simulator (Figure 1). The task consisted of a 120-kilometer drive on a motorway, beginning at one motorway rest area and ending at a second motorway rest area. Four scenarios inducing speed adjustments were presented on the route: a toll, a bus accident in the left lane, a construction area inthe right lane and the presence of radar. Each individual participated in three sessions, one using a speed limiter in the vehicle, the second using a speed regulator in the vehicle, and the third without a device to regulate or limit speed (control condition). Each participant performed the test on three distinct days, with the order of the three conditions counterbalanced across subjects. This method serves to neutralize any learning effect or habituation to the driving simulator, since each experimental condition is experienced first, second and third by the same number of subjects. At the beginning of each condition, subjects completed the Karolinska Sleepiness Scale (KSS) questionnaire (see page 19) assessing their level of alertness. If participants had a score higher than 4 on the KSS, the session was deferred to a later date. 6/22

Figure 1: Top: Two examples of 3D objects modeled and integrated in the virtual environment of the driving simulator of the CI2N, a motorway rest area (left panel) and a toll (right panel). Center: External view of the simulator (left panel) and photography inside the simulator (right panel). The simulator provides a virtual environment projected on screens spanning 360 degrees around the vehicle. Bottom: Left panel, a participant completes one of the tasks in the driving simulator. He is wearing a cap with 32 active electrodes recording, during the driving task, the electroencephalographic (EEG) oscillations resulting from the electrical activity of the brain. This device can detect oscillations sensitive to hypovigilance. Right panel, the control station where all the data can be viewed online. 7/22

4. Results All driving data collected during the three simulator sessions were derived from recordings of the simulator's parameters at a frequency of 10Hz. The results presented in this report include the factor "age". However, since statistical analyses of these data showed no effect of sex, no male/female distinction is reported here. 4.1. Analyses of driving on a simulator 4.1.1. Driving data related to road behavior 4.1.1.1. Average speed The average speed of the vehicle was analyzed over the entire route (i.e., during the entire 120-kilometer ride; Figure 2a), as well as by differentiating between the periods when the driver was traveling in the right (Figure 2b) and left (Figure 2c) lanes. We found that the average speed over the entire route was similar under the three driving conditions (F [2,261] =.16, p =.86), as were the average speeds when the car was traveling in the left, or passing, lane (F [2;261] =.47, p =.63) and the right lane (F [2;261] =.10, p =.90). We found that age had a significant effect on average speed over the entire route (F [2,261] = 8.0, p <.01). Post-hoc analyzes showed that subjects aged >60 years drove significantly more slowly than subjects aged 40-50 (p =.03) and 18-30 (p <.01) years, with the two latter groups having similar average speeds (p=.10). This effect of age was identical for all three driving conditions (i.e., control, limiter and regulator; F [4;261] = 4.47, p =.76). The average speed while traveling in the right lane differed significantly in the three age groups (F [2,261] = 4.34, p =.01). Subjects aged >60 years drove significantly more slowly than subjects aged 40-50 and 18-30 years (p=.01), whereas the latter two groups had similar average speeds (p =.09). This effect of age on average speed in the right lane was similar for all three experimental conditions (F [4;261] =.60, p =.67). 8/22

The average speed in the passing (left) lane also differed among the three age groups (F [2,261] = 7.70, p <.01), being similar in subjects aged 40-50 and >60 years (p =.11) but significantly lower than in subjects aged 18-30 years (p <.01 and p =.04, respectively). This effect of age on average speed in the passing lane was similar under all three experimental conditions (F [4;261] =.17, p =.95). 140 135 Subjects aged 18-30 years Subjects aged 40-50 years Subjects aged > 60 years Average vehicle speed (km/h) 130 125 120 115 110 Control Limiter Regulator Figure 2a: Average vehicle speed over the entire route for the three experimental conditions and according to driver age. Average speed of the vehicles when driving in the right lane(km/h) 140 135 130 125 120 115 110 Subjects aged 18-30 years Subjects aged 40-50 years Subjects aged > 60 years Control Limiter Regulator Figure 2b: Average speed of the vehicles when driving in the right lane for the three experimental conditions and according to driver age. Average speed of the vehicles when driving in the left lane (km/h) 140 135 130 125 120 115 110 Subjects aged 18-30 years Subjects aged 40-50 years Subjects aged > 60 years Control Limiter Regulator Figure 2c: Average speed of the vehicles when driving in the left (passing) lane for the three experimental conditions and according to driver age. 9/22

4.1.1.2. Overtaking and pulling back For each subject, we calculated the number of lane changes made to overtake a vehicle during the 120-kilometer ride. A lane change corresponded to the vehicle going from the right lane to the left lane, a maneuver performed to pass another vehicle. We found that driving condition had no effect on the mean number of lane changes (F [2,261] =.10, p =.91; Figure 3a), but did affect vehicle speed during lane changes (F [2,261] = 3.99, p =.02). Post-hoc analyses showed that the average speed during a lane change was significantly higher under control than under "limiter" (p=.02) and "regulator" (p=.05) conditions (Figure 3b), but was similar in the latter two conditions (p=.39). We further analyzed the average inter-vehicle distances during lane changes and when returning to the right lane after passing. Distances were calculated between the participant's vehicle and the car ahead of it while changing lanes and between the participants vehicle and the car behind when returning to the right lane (Figure 4). Although we found an average decrease of 5% in inter-vehicle distances during lane changes for both speed regulator conditions compared to the control condition, these distances did not differ statistically in the three experimental conditions (F [2,261] =.24, p =.79). In contrast, the distance between the subject's vehicle and the vehicle just passed when returning to the right lane differed significantly among the three experimental conditions (F [2,261] = 2.96, p =.05). Post-hoc analyses showed that intervehicle distances were shorter when using a limiter or regulator than under the control condition (p =.05 each), with an average decrease of 10%, although these distances were similar for the former two conditions (p =.73). The driver's behavior when passing or returning to the right lane was also influenced by age. Indeed, we found that age significantly affected the number of passing incidents (F [2,261] = 8.67, p <.01) and theaverage passing speed (F [2, 261] = 6.00, p <.01) during the entire trip. Post-hoc analyses showed that the number of passing incidents by drivers aged >60 years was significantly lower than those by drivers aged 40-50 (p =.03) and 18-30 (p <.01) years and was significantly higher by subjects aged 18-30 than 40-50 years (p =.05). Average passing speeds by drivers aged >60 and 40-50 years were similar (p =.27), but were significantly slower than those of subjects aged 18-30 years (p <.01 and p =.02, respectively). These effects of driver age on the number of passing incidents (F [4,261] =.44, p =.78) and average passing speed (F [4,261] =.10, p =.97) were equivalent for the three experimental conditions (i.e., control, speed limiter and speed regulator). Age 10/22

did not significantly affect inter-vehicle distances while passing (F [2;261] = 1.62 ; p =.20) and returning to the right lane (F [2;261] =.13 ; p =.88). 15 Subjects aged 18-30 years Subjects aged 40-50 years Subjects aged > 60 years 14 number of lane changes for passing 13 12 11 10 Control Limiter Regulator Figure 3a: Mean number of lane changes (for passing) performed during the entire 120-kilometer drive under the three conditions and according to driver age. Average passing speed during lane changes (km/h) 136 135 134 133 132 131 130 129 Subjects aged 18-30 years Subjects aged 40-50 years Subjects aged > 60 years 128 Control Limiter Regulator Figure 3b: Average passing speed during lane changes for the three experimental conditions and according to driver age. 90 50 Mean inter-vehicle distances during lane changes (m) 85 80 75 70 65 60 55 71,77 69,00 69,26 Mean inter-vehicle distances during passing (m) 48 46 44 42 40 38 36 34 32 42,52 71,77 38,32 39,04 50 Control Limiter Regulator 30 Control Limiter Regulator Figure 4: Mean inter-vehicle distances during lane changes (left panel) and passing (right panel) for the three experimental conditions. 11/22

4.1.2. Driving parameters sensitive to hypovigilance The parameters analyzed here are those sensitive to drowsiness while driving. -Adjustment of the lateral position of the vehicle, or maintaining the vehicle's path, is particularly sensitive to the extent of drowsiness. Reduced alertness leads to higher amplitudes and lower frequencies of adjustments 1. -When a driver becomes drowsy, adjustments in speed increase in amplitude, while the number of adjustments decreases. However, this variable was not relevant for the present study since speed was controlled by speed regulation and limitation systems. Hence, this parameter will not be considered here as a measure of drowsiness 2. 4.1.2.1. Lateral position and speed adjustments For each subject and for the three conditions, we calculated the distance of the vehicle relative to the central axis of the motorway during the entire ride (at a frequency of 10Hz). From these data, we calculated the mean amplitude of adjustments of the lateral position during the 120-kilometer ride, as well as the frequency of these adjustments, expressed as number per minute. Variations lower than 0.1m were excluded from analysis, as they were considered insignificant. The amplitude of adjustments of the lateral position (F [2,261] = 5.12, p <.01), as well as the frequency of these variations (F [2,261] = 4.11, p =.02), differed significantly among the three experimental conditions (Figure 5). Post-hoc analyses showed that the amplitude of adjustments was similar when using a limiter or regulator (p=.54), but was significantly greater when drivers used a speed limiter (p=.02) or regulator (p <.01) than under the control condition. Compared to the control condition, the amplitude was 22% and 33% higher for the limiter and regulator conditions, respectively. The frequency of adjustments of the lateral position was significantly greater under control than under limiter (p =.01) and regulator (p =.05) conditions, with frequency similar under the two conditions of automated speed regulation (p =.29). On average, when using the 1 Verster JC, Roth T. Vigilance decrement during the on-the-road driving tests: The importance of time-on-task in psychopharmacological research. AccidAnalPrev. 2013 ; 58:244-8. 2 Howard ME, Jackson ML, Kennedy GA, Swann P, Barnes M, Pierce RJ. The interactive effects of extended wakefulness and low-dose alcohol on simulated driving and vigilance. Sleep. 2007; 30(10):1334-40. 12/22

regulator or limiter, drivers readjusted the position of the vehicle 25% less frequently than under the control condition. We found that the amplitude of adjustments of the lateral position was equivalent among the 3 age groups (F [2, 261] =.11, p =.90), regardless of experimental condition (F [4,261] =.58, p =.68). In addition, age did not significantly affect the frequency of adjustments of the lateral position (F [2,261] =.29, p <.74), regardless of the experimental condition (F [4;261] =.44 ; p =.78). Mean frequency of adjustments of the distance of the vehicle relative to the central axis of the motorway (n/min) 1,10 1,05 1,00 0,95 0,90 0,85 0,80 0,75 0,70 0,65 0,60 0,55 0,50 0,91 0,87 0,80 Control Limiter Regulator Mean amplitude of adjustments of the distance of the vehicle relative to the central axis of the motorway (cm) 22,0 21,5 21,0 20,5 20,0 19,5 19,0 18,5 18,0 20,88 19,78 19,28 Control Limiter Regulator Figure 5: Amplitude (left panel) and frequency (right panel) of adjustments of the distance of the vehicle relative to the central axis of the motorway in the three experimental conditions. 13/22

4.1.2.2. Deceleration distance at the occurrence of a scenario Four scenarios prompting the drivers to adjust their speed punctuated the route: a toll, a bus accident in the left lane, a construction area in the right lane and the presence of radar. For each of these situations, we calculated the distance from the event at which subjects first pressed the brake pedal (Figure 6). We found that the distance at which subjects decelerated differed significantly among the experimental conditions (F [2,261] = 7.22, p <.01). Post-hoc analyses revealed that these distances were significantly lower when drivers were using a speed limiter or regulator than under the control condition (p=.01 each), but that the distances were similar when using either of the two speed regulation systems (p =.20), corresponding to an additional 40 meters traveled before pressing the brake pedal compared to the control condition. Driver age, however, did not affect the distance between first braking of the vehicle and the event (F [2;261] =.47 ; p =.62). Mean distances of a vehicle at which the driver pressed the brake pedal when approaching a risk event (m) 350 325 300 275 250 225 200 175 150 Control Limiter Regulator Subjects aged 18-30 years Subjects aged 40-50 years Subjects aged >60 years Figure 6: Mean distances of a vehicle at which the driver pressed the brake pedal when approaching a risk event under the three experimental conditions and according to the driver age. To refine our results, we integrated the factor "scenario" into the analysis of braking distance (Figure 7). By integrating this factor, we were able to analyze whether the distances at first pressing the brake differed for the four risk events: Construction area in the right lane at kilometer 46. Slowing of traffic due to the presence of radar at kilometer 68. Bus accident in the left lane at kilometer 90. Slowing of traffic due to a toll at kilometer 111. 14/22

Analyses showed that, when all groups were combined, the deceleration distance at the occurrence of a risk event differed among these events (F [3,351] = 7.22, p <.01). Post-hoc comparisons showed that the distance at which the brake pedal was first pressed was significantly greater for scenario 1 (construction area) than for the other three (p <.01). Distances measured for scenarios 2 (radar) and 3 (bus accident) were equivalent (p =.70), whereas the distance for scenario 4 (toll booth) was significantly shorter than for scenario 1 (p <.01) and scenarios 2 and 3 (p =.05 each). Thus, while driving, our participants, whether or not using a speed regulation system, braked progressively later at the approach of a risk event. Although analyses showed no interaction between the factors "scenario" and "experimental condition" (F [6,531] =.62, p =.71), analyses of each scenario showed that, for all four situations, the use of a speed limiter or regulator significantly decreased the braking distance while approaching a risk event compared to the control condition (p<.01 each). Although these braking distances were similar whether a speed regulator or a speed limiter is used for scenarios 1, 2 and 3, the distance was significantly lower for scenario 4 when using a regulator than when using a speed limiter (p <.01). Mean distance of a vehicle at which the driver pressed the brake pedal when approaching each of the four risk events (m) 400 350 300 250 200 150 100 50 282,40 238,52 209,95 229,31 183,76 161,84 211,36 202,95 174,76 174,04 172,15 117,51 Control Limiter Regulator Scenario 1 (km 46) Scenario 3 (km 90) Scenario 2 (km 68) Scenario 4 (km 111) Figure 7 : Mean distance of a vehicle at which the driver pressed the brake pedal when approaching each of the four risk events under the three experimental conditions. These results suggest that the deleterious effects on driving performance of using a speed regulator and speed limiter increase over time, with more pronounced effects in scenario 4 than in scenario 1. This observation led to an analysis of all driving parameters measured under the three conditions during four time periods while driving: 0-14, 15-30, 31-45, and 45-60 minutes. 15/22

Two parameters varied differently over time, depending on whether or not the driver was using a speed limiter or regulator: the mean amplitude (F [3,351] = 8.47, p <.01) and the mean frequency (F [3,351] = 5.62, p <.01) of the distance of the vehicle from the central axis of the motorway (Figure 8). This variation interacted statistically with the three experimental conditions (F [6,531] = 2.28, p =.04). Although these two parameters did not differ significantly among the three experimental conditions over the first 45 minutes of simulated driving, the mean frequency of adjustments of the lateral position was significantly higher and the mean amplitude of adjustments significantly lower during the last 15 minutes when using the regulator than when using the limiter or in the "scenario" condition (p <.01 each). During this period, the amplitude measured in the "control" condition was also significantly higher than that observed in the "control" condition (p <.01). Mean amplitude of the distance of the vehicle from the central line of the motorway (cm) 30 29 28 27 26 25 24 23 22 21 20 19 18 17 16 15 Control Limiter Regulator 0-14 min 15-30 min 31-45 min 45-60 min Mean frequency of the distance of the vehicle from the central line of the motorway (n/min) 1,20 1,15 1,10 1,05 1,00 0,95 0,90 0,85 0,80 0,75 0,70 0,65 0,60 0,55 0,50 Control Limiter Regulator 0-14 min 15-30 min 31-45 min 45-60 min Figure 8: Mean amplitude and frequency of the distance of the vehicle from the central line of the motorway in the three experimental conditions and during the four 15 minute periods of the simulated driving task. 16/22

4.1.3. Summary of analyses of driving data Table 1 summarizes our analyses of the driving data. Table 1: Summary of analyses of data on driving. Driving variables "Limiter" condition "Regulator" condition Between the different age categories Average speed over the entire trip = = Average speed in the right lane = = Average speed in the left lane = = Mean number of lane changes = = Average speed during lane changes Mean inter-vehicle distance when passing Mean inter-vehicle distance when returning to the right lane Amplitude of DVA adjustments Mean frequency of DVA adjustments Mean distance increase between vehicle and risk event when brake pedal is first pushed - 2 km/h max observed: - 10 km/h - 2 km/h max observed: - 10 km/h S < F S <Y F = Y S <F S < Y F = Y S = F S < Y F<Y S < F S < Y F < Y S = F S < Y F < Y = = S = F = Y - 4m max observed: - 26 m + 2cm max observed: + 7 cm - 4% max observed: - 24% + 60m max observed: + 132 m - 4m max observed: - 28 m + 2cm max observed: + 9 cm - 9% max observed: - 24% + 60m max observed: + 158 m S = F = Y S = F = Y S > F S > Y F = Y S = F = Y Each variable in the limiter and regulator condition is compared with that in the control condition, with = indicating no difference between the speed control system and the control condition indicating a lower value for the speed control system than for the control condition + indicating a higher value for the speed control system than for the control condition Significant differences between either speed adjustment condition and the control condition is indicated by a bluebox. The last column shows the results for the three age categories: (S)eniors (F)orties and (Y)oung. DVA: distance of the vehicle from the central axis 17/22

4.2. Analysis of the physiological indicators of vigilance Three indicators were used to measure the state of alertness of the participants: selfassessment, using the Karolinska Sleepiness Scale (KSS) questionnaire; the spectral power of alpha oscillations; and eye movements. 4.2.1. Self-evaluation of the level of vigilance Participants were asked to complete the KSS questionnaire 5 times: just before departure from the first rest area (time T 0 ) and every 15 minutes during the 120-kilometer ride. The questionnaire, scored on a 9-point scale, was arranged into the following categories: 1 very much awake 2 very awake 3 awake 4 rather awake 5 neither awake, nor asleep 6 a little asleep 7 asleep, but able to react 8 asleep, and little able to react 9 very asleep, very little able to react, fighting against sleep Over time, the level of vigilance decreased, to an average between 4 ("rather awake") and 5 ("neither awake, nor asleep") (Figure 9). This reduction in vigilance increased significantly after 30 minutes of driving in the three experimental conditions. Increased drowsiness was significantly higher when using a speed regulator than under the control condition. In contrast, use of aspeed limiter did not result in increased fatigue compared with the control condition. The effect of the speed regulator was particularly marked in drivers aged 18-30 years (Figure 10), but less so in participants aged 40-50 and >60 years. 18/22

6,0 5,5 Control Regulator Limiter 5,0 4,5 4,0 KSS 3,5 3,0 2,5 2,0 1,5 1,0 0 15 30 45 60 Times (Min) Figure 9 : Mean self-assessment of vigilance and sleepiness on the KSS. 8 7 Control Regulator Limiter 6 5 KSS 4 3 2 1 0 0 15 30 45 60 0 15 30 45 60 0 15 30 45 60 1-30 years 40-50 years > 60 years Figure 10 : Mean self-assessment of vigilance and sleepiness on the KSS for the three age categories. 4.2.2. Alpha oscillations measured by electroencephalography Alpha rhythm refers to a cerebral rhythm, that is to say, an electroencephalographic (EEG) oscillation, resulting from the electrical activity of the brain, whose frequency is between 8 and 12Hz. Alpha rhythm occurs when an awake person closes his/her eyes and relaxes, or when he/she enters a state of drowsiness and/or sleepiness. Figure 11 shows an example of an alpha rhythm on an EEG (8-12 oscillations of high amplitude per second). The "amount" of alpha waves, called spectral power, can be extracted from the time domain to the frequency domain of the EEG by means of FFT (Fast Fourier Transform). 19/22

Figure 11: Example of an alpha rhythm indicating a decrease in alertness. The signal was extracted from the EEG by FFT. Figure 12 shows the spectral power of the alpha rhythm as a function of time. The total ride of 60 minutes was divided into periods of 10 minutes each to assess changes in alpha rhythm over time. 1200 1100 Control Limiter Regulator Mean Alpha spectral power (uv²/hz) 1000 900 800 700 600 Scenario 2 Scenario 3 Scenario 4 500 400 0 10 20 30 40 50 60 Times (min) Figure 12: Changes in the amount of alpha waves during the 60 minutes of driving. Analysis of the alpha rhythm confirmed the results of the KSS, showing that periods of drowsiness were increased when using a speed regulator than under the two other conditions. Alpha rhythm analysis also showed that use of a speed limiter resulted in greater drowsiness while driving than in the control condition, although to a lesser degree than was observed for the speed regulator, a finding not observed using the KSS. The increase in the amount of alpha waves was highest after one hour of driving. The increases compared to the control condition were 16% and 25% for the limiter and regulator conditions, respectively. 20/22

Interestingly, we found that the amount of alpha waves decreased during the four scenarios, reflecting the enhanced attention and actions by drivers required during these periods. Moreover, during these scenarios, the variations in alpha waves were larger when using a speed regulator than under the other two conditions, which may reflect the greater effort that must be expended to handle the regulator. In contrast to the speed limiter, the regulator must be disabled manually when the driver does not push the brake, making the regulator more difficult to manipulate than the limiter. 4.2.3. Ocular movements Ocular activity may also bean indicator of the level of driver arousal since this activity decreases when falling asleep. As a driver becomes more drowsy, his/her gaze becomes fixed. We therefore utilized electro-oculography (EOG) to measure eye movements while driving. Figure 13 illustrates vertical eye activity throughout the entire ride. 16000 15000 Control Regulator Limiter Ocular activity (vertical movements) (uv/ms) 14000 13000 12000 11000 10000 9000 8000 Scenario 2 Scenario 3 Scenario 4 7000 6000 0 10 20 30 40 50 60 Times (Min) Figure 13: Ocular activity (vertical movements) during the 60 minutes of driving. We found that vertical eye movements were less frequent when using the speed regulator than when using the speed limiter or under control conditions. This difference may be due to increased drowsiness when using the speed regulator, in agreement with the KSS and alpha rhythm results. However, vertical eye movements may also be reduced due to reduced need to check speed when using the regulator, and, to a lesser extent, the limiter. Indeed, the use of both speed control tools allows for less frequent visual checks of the dashboard. Moreover, we found that, as the participants approached the event points (toll, radar...), ocular activity increased under all three conditions. 21/22

5. Conclusion The results presented here clearly show that use of a speed regulator or limiter has significant effects on both driving behavior and driver vigilance. The effects on alertness were more marked when using a speed regulator than a speed limiter. We hypothesized that the lower motor activity resulting from the use of a speed regulator or limiter reduced vigilance while driving. Physiological measures, subjective sensations and driving parameters have confirmed this hypothesis, especially for the speed regulator. The speed limiter seems to have a lesser effect on alertness, but an effect as pronounced as that of the speed regulator on driving behavior. Self-assessment of the level of fatigue and alertness showed that use of the speed regulator resulted in more pronounced increases in fatigue and drowsiness after 30 minutes of driving. Interestingly, we found that drivers in their forties and seniors felt less fatigued than younger drivers and were less affected by the use of a speed regulator. These self-assessment results were confirmed by analyses of physiological indicators (alpha rhythm in EEG and eye movements). In contrast, statistical analyses of driving indicators sensitive to drowsiness showed that age had no effect on these parameters. Driving behaviors are also affected by the use of a speed regulator. Indeed, drivers tended to shorten their distance from the vehicle ahead when using the speed regulator, both in areas of moving traffic and in areas of high traffic density, especially when approaching risk areas. The speed limiter has much more limited effects on alertness than the speed regulator. There were no significant differences in vigilance, whether by subjective self-assessments or alpha rhythms, between the limiter and control conditions. The reduced frequency of vertical eye movements observed when using the limiter may simply reflect a reduction in the verification of speed on the dashboard. Driving behaviors are nevertheless similarly modified by the speed limiter and the speed regulator, with decreases in inter-vehicle distances and lowered slow-down latencies. Finally, it should be noted that the effect of the speed regulator on vigilance was assessed in participants with high levels of vigilance before starting the test, and that they drove only for one hour. Therefore, it is likely that the observed effects would be even more pronounced for longer driving times and/or in situations where the level of fatigue is higher at the beginning of the ride. 22/22