Assisted and Autonomous Driving on Driving Simulators: Advanced features to make Dynamic Driving Simulators suitable for studies on Assisted and Autonomous Driving

摘要

With the upcoming diffusion of autonomous vehicles, a substantial modification of the human role in the driving action is taking place. While most of the effort has been put on making the car capable of safely moving in a complex environment, the human in the control loop is becoming a critical problem: on the one hand, the driver attention in driving actions is necessary to guarantee safety in any conditions, on the other hand the driver comfort has to be considered to make the driving experience satisfactory. In this paper balancing between these two aspects is effectively investigated by means of dynamic driving simulators, particularly addressing the impact of vehicle dynamic setup. The target user for these applications is a non-professional driver, which does not easily fit into the virtual environment. Proprietary Active Seat (AS) and Active Belts (AB) technologies are used in the driving simulator to reduce the gap between real and virtual environment. Advanced Multi-Sensory Motion Cueing Algorithm based on Nonlinear Model Predictive Control technique is used to coordinate somatosensory stimulation (AS/AB) with the vestibular one (Motion Cues). Also, a specific feature has been introduced in the Motion Cueing Algorithm to provide the driver with a realistic perception of the slip dynamics of the vehicle. The combination of the AS, AB and slip dynamics makes the driver capable of understanding the vehicle behaviour also during passive driving situations that possibly exclude visual information. Moreover, cognitive aspects have to be considered to collect information regarding driver attention and comfort, which are not retrievable by motion analysis. To this aim, an advanced biotelemetry device is used to collect driver's Heart Rate Variability and Skin Potential Response, which are recognized bio-signal for stress load, and that are processed by means of statistical tools to infer cognitive features. Finally, practical use cases will be presented analysing the effects of different vehicle setups on human perception while experiencing autonomous driving on the simulator.