In-Vehicle Embedded Control Systems for Safety: a Survey

The design of a control system embedded in the vehicle to help the driver in the case a potential danger appears on the road – called also a driver assistance system - is a very complex problem, and needs knowledge from different domains. Generally speaking, an active driver assistance system can be defined as follows: if a calculated risk is bigger than a threshold (where the risk has also to be defined and this definition is itself a challenge), a control system acting on one or more actuators is activated to bring the vehicle back to a safe state. The control is deactivated when the risk is smaller than the threshold. However, rather than searching only a solution to a control problem, a set of very different elements must be integrated to the control design for the achievement of an effective system. For example, the following items, that are in addition often very closely linked between them, need to be considered : i) the traffic scenario on which the accidental situation to be handled is situated; ii) “the time before the accident” on which the system will be activated (also called time-to-the-accident); iii) the duration of the control system activation; iv) the reliability of the vehicle localization inputs to the control (lane markers detection, obstacle detection…); v) the inputs from human-machine-cooperation (HMC) studies, for example the positive and negative interferences between the driver and the machine; vi) if an open-loop or a closed loop control should be used; to name a few. The main contribution of the paper is then to make a review of a series of parameters and different elements, as point out, that need to be sum up and integrated one to the other for addressing the complex task of helping by active control a human-being when he is driving.