Information Technology Reference
In-Depth Information
actions. Evidently, it will be a matter of accompanying human activity as well as
possible, when it is relevant (i.e. compatible with the objectives of comfort,
performance and security to which the driver and/or human-machine system is
aspiring to). A copilot is certainly not meant to maintain the drivers in error and
back up their decision if they make an obvious mistake. On the contrary, it will be a
matter of making them aware of the danger, to try and persuade them not to engage
in a situation in which they will be unable to appreciate the critical nature, or to
assist them in mastering risks via an adequate corrective action.
Finally, the development of an automobile copilot invites us to go beyond the
strict frame of the “human-machine interaction” to inscribe the human-machine
relation in the context of a true “cooperation”. It is indeed the human-machine
system as a whole that is in charge of driving, driver and assistants joining their
efforts and skills in order to reach a common objective. This requires the abilities of
observation, interpretation and cooperation on the behalf of the copilot. A successful
cooperation can pass directly in the sensorimotor activities. Great progress has been
made in modeling these processes, but very delicate questions arise when we wish
to act with the best efficiency and at the best time. Furthermore, since the driver is
most able to provide a rich analysis of the situation, any interventions of the
automatic devices must be able to be modulated by this analysis. All the cooperation
modes pose the same problem of sharing a common frame of reference between the
driver and the automaton. Differences that are too large between the risk levels
evaluated by the two agents can put the efficiency of assistance devices in jeopardy.
In particular, the incoherence between information that is perceived or received by
the driver and the interventions by the device considerably affects confidence
[LEE 04]. Consequently, a key issue that will need to be dealt with more directly is
that of communication between the driver and the automatic devices, whether this
communication is based on information of a symbolic nature (vocal messages or
visual pictograms) or on a sub-symbolic mode (vibration of the steering wheel or
hardening of the pedals, for example). It is thanks to this communication that the
common frame of reference will regularly be able to be updated, all the while
looking to reduce cognitive costs for the human driver.
The three fundamental principles that should rule over an automobile copilot are
“integrative”, “adaptive” and “cooperative”. On the basis of this analysis, it is
possible to propose a general architecture of the copiloting device adapted to driving
of the vehicle (see Figure 5.6).
The copilot inputs correspond to the data coming from:
- the road environment (i.e. from the embedded perceptive sensors or directly
transmitted by the infrastructure, other vehicles or by a control post tasked with
regulating traffic);
Search WWH ::




Custom Search