Vehicle Dynamics Control after Impacts in Multiple-Event Accidents
Accidents statistics show that multiple-event accidents (MEAs) represent a considerable and increasing proportion of all vehicle traffic accidents. MEAs are characterized by having at least one vehicle subjected to more than one harmful event. MEAs now comprise approximately 25% of all passenger vehicle accidents. This thesis aims to make systematic progress towards developing a vehicle Post Impact Control (PIC) function so as to avoid or mitigate any secondary event in MEAs.
To characterize the vehicle motion control problems for PIC, a number of MEAs from an accident database were analyzed. Post impact vehicle dynamics were studied considering the overall accident scenarios of exemplar cases. Reduction of kinetic energy and path lateral deviation were found to be most critical and beneficial for the vehicles after impacts. To understand the mechanism of influencing the post impact vehicle path, numerical optimization was applied to minimize the maximum path lateral deviation. It was found that effective control can be achieved across a wide range of kinematic conditions, by switching between three sub-strategies established at vehicle body level. Results also showed that active front-axle steering, in addition to individual-wheel braking, provides significant control benefits, although not for all post-impact kinematics.
For closed-loop design of the path control, a Quasi-Linear Optimal Controller (QLOC) was proposed and verified with the numerical optimization results. The design method is novel – it well combines the linear co-states dynamics and nonlinear constraints due to tyre friction limits. The algorithm was further adapted to penalize both longitudinal and lateral path deviations, using a generalized cost function.
To verify the function with driver interaction, a number of exploratory methods were investigated regarding the driver safety, as well as the capability and accuracy to reproduce the real-world post-impact vehicle kinematics. A scheme of the function design for real-time implementation was proposed and applied to the experiments in a driving simulator environment.