Figure 1.
Density distributions as measured in the environment (
in red,
in blue, axes chosen in such a way that walking on the left corresponds to have
on the left of the figure). Error bars are obtained as standard deviations of values of
averaged over time windows of length 1200 s.
Figure 2.
Velocity distributions as measured in the environment (
in red,
in blue). Error bars are obtained as standard deviations of values of
averaged over time windows of length 1200 s.
Figure 3.
Relative velocity field around a pedestrian (based on all available observational data in the data collection location).
The frame is centred on pedestrian , the
axis gives the relative position of pedestrian
in the direction of
(i.e.,
), while the
axis gives the relative position from left to right. The colour-bar reports average
values in mm/s. The difference between values on the left and right, along with the very high relative velocity (violet) area on the back-right, suggest the presence of a right-oriented overtaking norm. The low velocity on the back (high velocity on the front) is probably related to collision avoiding behaviour for pedestrians moving in the same direction, while the low velocity area on the right and high velocity area on the left are probably related to group behaviour (friends that find themselves ahead will slow down and vice versa).
Figure 4.
Collision avoiding and overtaking in pedestrian models using the TP (Tilt in Position) condition.
Figure 5.
Collision avoiding and overtaking in pedestrian models using the TV (Tilt in Velocity) condition.
Figure 6.
Difference between avoiding a collision and being overtook in pedestrian models using the TV (Tilt in Velocity) condition.
A: since , i.e.
,
expects
to avoid on the left. B: since
, i.e.
,
expects
to overtake on the right.
Table 1.
Models and conditions performance ().
Table 2.
Models and conditions performance ().
Table 3.
Models and conditions performance on (
).
Table 4.
Models and conditions performance on (
).
Figure 7.
Comparison between the observed density distribution (black) and simulated density distributions
(blue and red) in
.
Both simulated distributions are obtained using the ES (Elliptical Specification) model, the blue line showing results obtained under the TP (Tilt in Position) condition, the red line results obtained under the TV (Tilt in Velocity) condition. Average values and standard deviations (error bars) for simulated distributions are obtained over tests (using
) of the overall best solution over the
independent runs of the GA (calibration over all the environments). Observed data error bars are obtained as in Fig. 1.
Figure 8.
Comparison between the observed velocity distribution (black) and simulated velocity distributions
(blue and red) in
.
Both simulated distributions are obtained using the ES (Elliptical Specification) model, the blue line showing results obtained under the TP (Tilt in Position) condition, the red line results obtained under the TV (Tilt in Velocity) condition. Average values and standard deviations (error bars) for simulated distributions are obtained over tests (using
) of the overall best solution over the
independent runs of the GA (calibration over all the environments). Observed data error bars are obtained as in Fig. 2.
Table 5.
Parameter ranges in models and conditions.
Table 6.
Parameter values after calibration in the ES (Elliptical Specification) model.
Table 7.
Parameter values after calibration in the CP (Collision Prediction) model.