4.1 Evaluation of the personal data
The average age across all test groups is 41 years. The average driving experience is 22.5 years. 20% of the test persons have less than 5000 km driving experience per year. 28.6% of the test persons drive between 5000 and 10,000 km per year. 30.0% of the test persons drive between 10,000 and 20,000 km per year. 21.4% of the test persons drive more than 20,000 km per year. 68.1% of the test persons have no driving simulator experience.
4.2 Evaluation of the questionnaires
The global assessment of the adaptive high-end system does not contain any very bad ratings (Fig. 6). 12% of the test persons rated the system as a whole rather bad, 52.3% of the test persons rated rather good and 35.7% of the test persons rated the global assessment of the system very good.
The global assessment of the adaptive system contains 2.6% very bad ratings. 28.6% of the test persons rated the system as a whole rather bad, 48.6% rated rather good and 20.3% rated the global assessment of the system very good.
The global assessment of the non-adaptive system contains 4.8% very bad ratings. 24.9% of the ratings were ratrher bad for the system as a whole. 47.5% rated the system rather good and 22.8% of the test persons rated very good. Two-tailed Wilcoxon test showed no significance at the level p < 0.05.
The HUD warning of the adaptive high-end system was rated very bad by 4.5% of the test persons. 11.8% of the test persons rated rather bad, 16.8% rated rather good and 66.8% rated very good.
18.4% of the test persons rated the optical warning of the adaptive system very bad. 50% of the test persons rated rather bad, 18.4% rated rather good and 13.2% rated very good.
The optical warning of the non-adaptive system was rated very bad by 34.9% of the test persons, 23.4% rated rather bad and 28% rated rather good. 13.7% of the test persons rated the optical warning of the non-adaptive system very good. Two-tailed Wilcoxon test showed significance at the level p < 0.05 with a p-value of 0.00398. Therefore, the result is even significant at p < 0.01.
The acoustic warning of the adaptive high-end system was rated very bad by 5.1% of the test persons. 15.5% of the test persons rated rather bad, 35.5% of the test persons rated rather good and 43.6% rated very good.
The acoustic warning of the adaptive system was rated very bad by 3.1%, 12.5% rated rather bad, 41.3% rated rather good and 43.1% of the test persons rated very good.
The acoustic warning of the non-adaptive system was rated very bad by 4.8% of the test persons. 14% of the test persons rated rather bad, 40.6% rated rather good and 40.6% rated very good. Two-tailed Wilcoxon test showed no significance at the level p < 0.05.
The brake pulse of the adaptive high-end system was rated very bad by 2.4%. 34.1% rated rather bad, 55.6% of the test persons rated rather good and 7.9% rated very good.
The brake pulse of the adaptive system was rated very bad by 5.9% of the test persons, 22.7% rated rather bad, 54.4% rated rather good and 17% rated the brake pulse of the adaptive system very good.
The brake pulse of the non-adaptive system was rated very bad by 7.7%. 29.7% of the test persons rated rather bad, 48.3% rated rather good and 14.3% of the test persons rated very good. Two-tailed Wilcoxon test showed no significance at the level p < 0.05.
The partial deceleration of the adaptive high-end system was rated rather bad by 25% of the test persons. 50% of the test persons rated rather, 20% rated very good. None of the test persons rated very bad.
The partial deceleration of the adaptive system was rated very bad by 5.3%. 18.9% rated rather bad, 45.9% rated rather good and 30% rated the partial deceleration of the adaptive system very good.
The partial deceleration of the non-adaptive system was rated very bad by 4.8% of the test persons. 18.5% of the test persons rated rather bad, 48.8% rated rather good and 27.9% rated very good. Two-tailed Wilcoxon test showed no significance at the level p < 0.05.
The global assessment of the adaptive high-end system does not contain any very bad ratings (Fig. 6). 50.9% of the test persons rated the system as a whole rather bad. 44.7% rated the rather good and 4.3% rated the global assessment of the adaptive high-end system very good.
The global assessment of the adaptive system contains 2.1% very bad ratings. 37.5% of the test persons rated the system as a whole rather bad, 50% rated rather good and 10.4% rated the global assessment of the system very good.
The global assessment of the non-adaptive system does not contain very bad ratings. 25% of the test persons rated the system as a whole rather bad, 60.4% rated the system rather good and 14.6% of the test persons rated very good. Two-tailed Wilcoxon test showed no significance at the level p < 0.05.
The HUD warning of the adaptive high-end system was rated very bad by 16.7%. 41.7% of the test persons rated rather bad, 22.9% rated rather good and 18.8% rated very good.
The optical warning of the adaptive system was rated very bad by 13.1% of the test persons. 44.5% rated rather bad, 33.4% rated rather good and 9% of the test persons rated very good.
The optical warning of the non-adaptive system was rated very bad by 14.7%. 38.5% rated rather bad, 29.9% rated rather good and 16.9% of the test persons rated very good. Two-tailed Wilcoxon test showed no significance at the level p < 0.05.
The partial deceleration of the adaptive high-end system was rated very bad by 2.6% of the test persons. 20.4% of the test persons rated rather bad. 60.8% rated rather good and 16.2% of the test persons rated very good.
2.2% of the test persons rated the partial deceleration of the adaptive system very bad. 27.5% rated rather bad, 56.8% of the test persons rated rather good and 13.5% rated very good.
Figure 7 shows the ratings of the total impression, the optical warning and the partial deceleration intervention of the 3 system variants in attentive driving phases.
The partial deceleration of the non-adaptive system was rated rather bad by 21.8%. 50.2% rated rather good and 28% of the test persons rated very good. None of the test persons rated very bad in any case. Two-tailed Wilcoxon test showed no significance at the level p < 0.05.
The following figure (Fig. 8) shows the results in direct comparison, which system variant the test persons preferred. Two of the three system variants were shown during the comparison test drive. Each system was seen and evaluated by 48 test persons. In total, the ratings of 72 test persons could be evaluated for three system variants.
The non-adaptive, close-to-series system was preferred by 31% of the test subjects. The adaptive system was preferred by 19.7%. The adaptive high-end system was preferred by 49.3% of the test subjects. Two-tailed Wilcoxon test showed no significance at the level p < 0.05. The difference between the rating of the adaptive system and the high-end adaptive system just missed significance with a p-value of 0.0536.
The following Figs. 9 and 10 show the ratings of warning and intervention times. The attentive driving situation and the test phase under distraction were evaluated seperately. The test persons rating options distingusted between much too early, too early, just right, little too late and much too late. Figure 9 shows the evaluation of the rating for the attentive test drive phase.
Both the adaptive system and the adaptive high-end system was classified by 40% of the subjects as little too late in attentive driving situations with regard to the warning and intervention time. The warning and intervention time of the non-adaptive system was rated just right by almost 60% of the test subjects. The warning and intervention time of the adaptive system and the adaptive high-end system was rated just right by 32% of the test subjects. Individual ratings were in the areas much too early, little too early and much too late. These are similarly distributed and negligible.
Figure 10 shows the evaluation of the warning and intervention times of the three system variants under distraction.
The warning and intervention times of the adaptive high-end system were rated just right by almost 65% of the test persons. With about 20% less in this category, the non-adaptive system and the adaptive system were evaluated. The test drive of the non-adaptive system under distraction was rated most frequently little too late. Followed by the adaptive system with 30% and 20% for the adaptive high-end system. For all three system variants the ratings in the categories little too early and much too late are negligible compared to the categories just right and little too late. They are at a similar level. In none of the three system variants the warning and intervention times were rated much too early.
4.3 Evaluation of the CAN data
The safety aspect of the SDS system was verified by evaluating and comparing the warning and intervention frequency of all three system characteristics under distraction using CAN data.
The following Fig. 11 shows the results. The vertical axis displays the number of warnings and interventions per system characteristics, for each of the the individual warning modalities.
By far the fewest warnings outputs occur within the non-adaptive system variant. Both adaptive variants show four times more warnings and never go beyond the first warning level. After the visual-acoustic warning, all drivers reacted in time and braked themselves. The non-adaptive system triggered the brake pulse twice. The partial deceleration was activated once. Warnings or interventions were not measured in any of the system variants during attentive driving phases, since the subjects decelerated in time.