Bromsa eller styra : En stokastisk beslutsmodell för automatiska nödmanövrar för tunga fordon

Traffic safety is a global concern, with over 1.3 million fatalities annually resulting from road traffic crashes. Heavy vehicles such as trucks and buses pose a particular risk due to their size, weight, and potential hazardous cargo, which amplify the severity of collisions. To mitigate these risks, collision avoidance systems like Automatic Emergency Braking (AEB) and Automatic Emergency Steering (AES) are increasingly used. The choice between AEB and AES is often based on which maneuver can be initiated latest, allowing the driver more time to react independently. However, most existing decision models assume perfect perception of surrounding traffic, leaving the impact of state estimation uncertainties largely unexplored. To address this gap, a case study was conducted based on a realistic highway scenario involving an imminent forward collision risk for a heavy vehicle. State estimation data from real-world testing were collected and used to recreate the scenario in an offline simulation environment. Two main aspects were examined: the effect of uncertainty on time-based decision metrics, and the ability of a steering control system to track fifth- and seventh-degree polynomial trajectories for emergency steering. The results showed that AES decisions must be initiated earlier than AEB to avoid a collision under state estimation uncertainty. Additionally, factors such as available brake acceleration and lateral offset to the preceding vehicle significantly influenced intervention timing. The steering control system, which had previously demonstrated good performance for passenger vehicles, showed limitations in the heavy vehicle configuration. Although the planned trajectories respected the kinematic constraints on lateral acceleration and steering rate, these were exceeded in the simulation when the control system attempted to follow the planned paths. For future work, incorporating a friction estimator and state prediction is recommended to enhance the vehicle's situational awareness. Additionally, model predictive control (MPC) is proposed as an alternative steering control system for further investigation in the context of heavy vehicles.