Project summary

The project aims at contributing to make cars safer and more ergonomic through the integration of information and communication technologies, especially of driving assistance automation. It focuses on the safety function related to preventing involuntary lane departure, in bend or in straight line, on the basis of trajectory failures detection for prevention, of risky situations identification, and of human failures diagnosis, with the aim to intervene in anticipation of devices like ESP. It studies the cooperation between driver and automation, considering automation acceptability, driving styles (individual differences), and distractions. It pays a great attention to human-machine interfaces, especially sensorial interfaces, in order to reach an optimal efficiency, that is to say a desired effectiveness associated to a reasonable cognitive cost. The road departure function is related to 30% to 40% of the accidents, according to the year, the gravity or the country. Research has devoted less attention to this function than for others like speed or time headway control, although the function had been tackled within two generalist programs (ARCOS/PREDIT and PReVENT) and has been studied by a more specific program (PREVENSOR/PREDIT). Three cooperation modes have been defined in order to make easier trajectory failure recovery:enlightening the tangent point inside the bend); (ii) mutual control mode (auditory or haptic warning of departure from a safe trajectory, or action suggestion by motor priming from the steering wheel); and (iii) control mode (automatic lateral control). After eliminating the control mode for difficulties in returning to manual control and after identifying the limits of visual enhancement, it is now required to go deeper into mutual control research and to reach a more ambitious shared control mode. Within this concept, automation combines its action with the driver’s action in order to make the trajectory safer. Shared control can take the form of a limit mode where the drivers’ control is given a ceiling in order to avoid oversteering. It can also take the form of a corrective mode where a too weak or too strong driver’s control is corrected by automation in order to adapt to the situation before returning the control. Finally, and more originally, it can be a continuous shared control mode, where the control initiative can come from any side (e.g., automation can take the control in order to counter a wind gust. These new modes are real innovations and can open the way to “steer by wire" by a necessary upstream research. The PARTAGE project also aims at examining crucial human factors questions within this context: the individual and social acceptability of shared control, and means for improving them, consideration of risk factors more ecological than the simple road geometry (traffic and meaningful elements near the trajectory), and cognitive and cybernetic modelling of diverse driving styles. The project includes trajectory longitudinal and lateral control, especially by devoting attention to speed control when approaching a bend. The need for working on critical situations requires simulation. However, it is also required that proprioceptive sensations (e.g., accelerations) be reproduced in some experiments. That is why the project includes the design of a new test means: a virtual and augmented reality helmet on a real vehicle running on a test track, the validity of moving-base simulators remaining tenuous. The questions under study will be formulated at a generic level, in terms of human-machine cooperation,