Distance [m]
2.50
2.00
1.50
1.00
0.50
0.00
Clearance test Walking right of the
road marking (P-3) Walking on the road
marking (P-1) Walking left of the
road marking (P-2)
Figure 4.8: Lateral clearance between vehicle and pedestrian in the normal situations “passing”
and the clearance test. The circles in the boxplots indicate near outliers.
(i.e., the lowest lateral clearance of the pedestrian to the street) the drivers reacted 0.3 s earlier (SD 0.42 s to 0.62 s). The differences in the mean are non-significant, using t-tests (situations: 1 versus 2: t=−1.87; 2 versus 3: t=−0.56; 1 versus 3: t=−1.48).
The longitudinal reaction in the “passing” situations depended highly on the position of the pedestrian. A braking reaction is observed in
• 1 of 29 situations for condition P-3 (i.e., largest initial clearance);
• 7 of 35 situations for condition P-1 (i.e., middle initial clearance);
• 19 of 34 situations for condition P-2 (i.e., smallest initial clearance).
In order to have sufficient data for interpretation only condition P-2 was used for analysis of longitudinal driver reactions. The drivers started braking on the average at a TTC of 4.10 s (SD = 0.61 s).
The longitudinal behavior was further investigated using normal situation “crossing”.
The brake response in this situations is also dependent on the initial TTC of the situation itself.
• At the lowest TTC (approx. 5.4 s; condition C-1), 35 of 35 persons braked.
• At the medium TTC (approx. 6.6 s; condition C-2), 18 of 38 persons braked.
• At the largest TTC (approx. 7.8 s; condition C-3), 12 of 35 persons braked.
4.3 Acceptance of the system in specific situations
Average TTC [s]
5.00
4.00
3.00
2.00
1.00
0.00
Walking right of the road marking (P-3) Walking on the road
marking (P-1) Walking left of the
road marking (P-2)
Seite 1 Figure 4.9: TTC to the pedestrian at the beginning of the steering action in the normal
situations “passing”.
The initial reaction, i.e., taking the foot away from the accelerator pedal, was hard to evaluate, as many persons in this simulator experiment did not constantly apply the ac-celerator pedal. As a consequence, the results could be distorted by simulator artifacts and are therefore not discussed. The start of braking is a valid indicator for this situation and is shown in Fig. 4.10. The average TTC is 4.07 s (SD = 0.90 s) for condition C-1 and 3.82 s (SD = 1.65 s) for condition C-2. Due a low number of measurements (i.e., braking reactions), the results presented exclude condition C-3.
The results shown above allow for an interpretation concerning acceptance of possible system actions. The findings are consistent and show a stable lateral passing clearance around 1.5 m independent of the initial conditions or the test setting. The steering reaction is consistent and shows a reaction around 3 s TTC. Start of braking as longitudinal reaction was observed around 4 s TTC. As the situations were uncritical and the drivers had as much time to react as they liked (the pedestrians were visible long before the TTC values mentioned above), it can be concluded that these values indicate a comfort zone which the drivers like to maintain. It does not mean that they feel uncomfortable immediately below those values or that the situation is regarded as hazardous immediately below those values.
It can be concluded from the findings in the normal situations that a system configuration as described above will be accepted quite well. The system reaction (i.e., acoustical warning at a TTC of 1.5 s or automatic braking at a TTC of 0.9 s) is at a TTC where nearly all drivers would have reacted (using the results presented above) if they had the chance to perceive the pedestrian. This is also confirmed by the results of the interview regarding
Low TTC (C-1) Medium TTC (C-2)
TTC [s]
6.00
4.00
2.00
0.00
Figure 4.10: TTC to the pedestrian at the beginning of the braking reaction in the normal situations “crossing”.
the highly critical situation (see next section). It can also be concluded that at higher TTC levels than discussed using Fig. 4.10, nearly no drivers feel a necessity to react to the situation.
The next part focuses on the investigation of the acceptance situations mentioned above.
The research question is the subjective perception of hazard and acceptance of false system actions in these situations. The subjects were suddenly confronted with the situations.
Out of the 20 subjects, 10 got false warnings and an automatic braking (in case the TTC values became small enough) and the other 10 got only a warning without an automatic braking as system response. The system response was triggered in the situations to get a reliable presentation for as many subjects as possible. This part of the experiment has the characteristic of a presentation and is meant to produce qualitative insights into acceptance of false system actions, not to produce an amount of data that is statistically usable. A system ready to go into mass production would handle most of these situations by, e.g., predicting the trajectory of the vehicle, predicting the pedestrian’s movement or calculating a collision probability.
The first rating uses a 100 % scale to investigate the perceived hazard of the situation (where 100 means maximum hazard and 0 minimum hazard). The subjects were instructed to rate the hazard for thewhole surrounding traffic, not for the pedestrian alone. The rat-ing regardrat-ing hazard of the situation is displayed in terms of the median in Fig. 4.11, and the corresponding ranks are given in Table 4.1 for both conditions (i.e., warning only and full system response). Whereas the absolute value in this scale cannot be interpreted, the relative differences do have a meaning. The situations “curve”, “traffic island”, and
“inter-4.3 Acceptance of the system in specific situations
40 60 80 100
Median
0 20
Traffic island Evasive maneuver
Curve Intersection curve
T-intersection Parking bay
with automatic braking warning only
Figure 4.11: Rating “hazard of the situation”. Medians.
Table 4.1: Rating “hazard of the situation”. Ranks.
Traffic Evasive
Curve Intersection
T-intersection Parking
island maneuver curve bay
with automatic
2a 4 2 1 5 3
braking
warning only 1 3 6 4 5 2
a“Traffic island” shares the same rank with “Curve”.
section curve” were rated as most hazardous (followed closely by “parking bay”, “evasive maneuver”, and “T-intersection”). The subjects explained this by the unpredictability of the system reaction (“curve” and “intersection curve”) and by the current speed of the vehicle (“traffic island”). Comparing both conditions, the rating differs most for “curve”
and “intersection curve”. A possible explanation is that automatic braking irritates the subjects more while negotiating a curve than a straight road. It can be seen in every situation that automatic braking is regarded as more hazardous than only a warning (ex-ception: “T-intersection”).
The second rating focuses on acceptance of the undesired system response. It also uses a 100 % scale, where 100 stands for lowest acceptance and 0 for highest acceptance. As explained before, only relative differences do have a meaning. Fig. 4.12 gives the medians and Table 4.2 the ranks for the situations and both conditions. The lowest acceptance
40 60 80 100
Median
0 20
Traffic island Evasive maneuver
Curve Intersection curve
T-intersection Parking bay
with automatic braking warning only
Figure 4.12: Rating “acceptance of false system action”. Medians.
Table 4.2: Rating “acceptance of false system action”. Ranks.
Traffic Evasive
Curve Intersection
T-intersection Parking
island maneuver curve bay
with automatic
4 2 1 3 5 6
braking
warning only 5 3 1 4 2 6
can be observed in the situations “curve”, “evasive maneuver”, and “T-intersection”. The subjects provided two explanations for this:
• The pedestrian does not move and thus a real hazard, i.e., the need for a system action, cannot be seen.
• The pedestrian is on the far side of the street, and the resulting hazard for the pedestrian is also seen as low.
There are large differences between the conditionsfull system and warning only, especially in the situations “traffic island” and “parking bay”. Braking in front of the traffic island surprises and confuses the subjects. Automatic braking while passing the pedestrian next to the parking bay is also highly unacceptable compared to warning only, as the pedestrian has nearly been passed by the time the braking sets in.
The last acceptance situation (i.e., running pedestrian at intersection) has to be in-terpreted separately, as it does not present a false system action but a miss (i.e., false