Optimal Vehicle Motion Control to Mitigate Secondary Crashes after an Initial Impact.

Abstract

Statistical data of road traffic fatalities show that fatalities in multiple-event crashes are higher than in single-event crashes. Most vehicle safety systems were developed to mitigate first crash events. Few active safety systems can deal with subsequent crash events. After a first crash event, drivers may not react in a timely or correct manner, which can have devastating consequences. Production active safety systems such as Electronic Stability Control (ESC) may not react to a first crash event properly unless such events are within their design specifications. The goal of this thesis is to propose control strategies that bring the vehicle state back to regions where drivers and ESC can easily take over the control, so that the severity of possible subsequent (secondary) crashes can be reduced. Because the most contributing causes of fatal secondary crashes are large lateral deviations and heading angle changes, the proposed algorithms consider both lateral displacement and heading of the vehicle. To characterize the vehicle motion after a crash event, a collision force estimation method and a vehicle motion prediction scheme are proposed. The model-based algorithm uses sensing information from the early stage of a collision process, so that the collision force can be predicted and the desired vehicle state can be determined promptly. The final heading angles are determined off-line and results are stored in a look-up table for faster implementation. Linear Time Varying Model Predictive Control (LTV-MPC) method is used to obtain the control signals, with the key tire nonlinearities captured through linearization. This algorithm considers tire force constraints based on the combined-slip tire model. The computed high-level control signals are realized through a control allocation problem which maps vehicle motion commands to tire braking forces. For real-time implementation, a rule-based control strategy is obtained. Several rules were constructed, and results under the rule-based control are similar to those under the optimal control (LTV-MPC) method while avoiding heavy on-board computations. Lastly, this thesis proposes a preemptive steering control concept. By assessing the expected strength of an imminent collision force from another vehicle, a preemptive steering control is applied to mitigate the imminent impact.