Localisation-based autonomous vehicle rear-end collision avoidance by emergency steering

Abstract

Rear end collision is one hazard accident for an autonomous vehicle. Emergency braking has already been deeply researched and applied into some new model vehicles to avoid rear-end collision or mitigate its impacts. Emergency steering is another promising method to enhance vehicle safety; this kind of collision avoidance approach remains to be carefully studied. To fill this gap, this work proposes an emergency steering based rear-end collision avoidance approach concerning required safe distance as safety index. Vehicle dynamics, Tire Slip Angle and Segal lateral force model, are concerned to obtain the vehicle states. Monte Carlo simulation is executed to determine the value of required safe distance when encountering with a stationary and a suddenly decelerating target vehicle at different velocities. Safe Distance Margin and quantised influence two indexes are proposed to assess the influence of measurement uncertainty. Simulation and road test experiment results demonstrate that the proposed method is effective to avoid rear-end collision by emergency steering manoeuvre. Moreover, some aspects, like vehicle lateral stability, emergency manoeuvre selection and simultaneously emergency manoeuvres are discussed as well in this work in order to apply this collision avoidance method into practical application.