Information Technology Reference
In-Depth Information
As an example, a fall in collision time from 3.0 to 1.5 s during the arrival of a
slow vehicle does not mean the same thing depending on whether the driver: (i)
keeps the speed and trajectory stable; (ii) engages the brakes; or (iii) has activated
the left indicator. If, in the first case, the absence of reaction can potentially be
interpreted as a deficiency in the detection or estimation of the speed of the obstacle
(which could justify an alert), things are radically different in the two other
configurations. The action on the brake proves an awareness of the presence of an
obstacle, but nonetheless does not prejudge the efficiency of the engaged regulation
procedure (a breaking aid can be envisaged in this case). As to activation of the
indicator, it very probably reveals an intention to overtake, which it would be
unadvisable to counter by the diffusion of a collision avoidance alert, at the risk of
unnecessarily disturbing an overtaking maneuver, which in essence requires the
complete attention of the driver.
Evidently objective parameters, even as relevant as the time to collision or the
inter-vehicle distance, for example, are often insufficient to make the right decisions
in terms of human-machine interactions. It is precisely with the objective of
bringing these elements of decision into
real time
that the IFSTTAR laboratory of
ergonomics and cognitive science began (ARCOS
6
program supported by the
PREDIT
7
) the design and development of a device for the automatic analysis of the
driving situation. The ultimate ambition of this was to enable an adaptive
management of interactions between the driver and a collision avoidance device.
Summarily (see Figure 5.1) the objective is to enable a joint analysis of the
driver's activity (i.e. the actions that he engages on the commands and their effects
on the dynamics of the vehicle) according to the data enabling us to characterize the
driving situation (via information coming from perception technologies, as well as
mapping data), and this in order to:
-
Detect
situations that could potentially cause accidents (rapid approach
towards a slower vehicle, the arrival of a fixed obstacle on the road, the presence of
vulnerable users on the road, whether it be pedestrians or cyclists, etc.).
-
Analyze
, in real time, the behaviors of the driver with regards to objective
characteristics of the road situation (for example, how critical it is).
-
Assess
the adequacy of the behaviors implemented by the driver and diagnose
driving errors (for example, non detection of an obstacle).
6 ARCOS (2001-2004):
Action de Recherche pour une COnduite Sécurisée
(Research Action
for Safe Driving).
7 PREDIT:
Programme de REcherche et D'Innovation dans les Transports
terrestres
(Research and Innovation Program in Terrestrial Transport).
