This study proposes a novel collision avoidance and motion planning framework for connected and automated vehicles based on an improved velocity obstacle (VO) method. The controller framework consists of two parts, that is, collision avoidance method and motion planning algorithm. The VO algorithm is introduced to deduce the velocity conditions of a vehicle collision. A collision risk potential field (CRPF) is constructed to modify the collision area calculated by the VO algorithm. A vehicle dynamic model is presented to predict vehicle moving states and trajectories. A model predictive control (MPC)-based motion tracking controller is employed to plan collision-avoidance path according to the collision-free principles deduced by the modified VO method. Five simulation scenarios are designed and conducted to demonstrate the control maneuver of the proposed controller framework. The results show that the constructed CRPF can accurately represent the collision risk distribution of the vehicles with different attributes and motion states. The proposed framework can effectively handle the maneuver of obstacle avoidance, lane change, and emergency response. The controller framework also presents good performance to avoid crashes under different levels of collision risk strength.

It is of paramount importance for mitigating road crash losses to characterize the relationship between crash injury severities and contributing factors. Existing studies have revealed mechanism differences of single-vehicle (SV) and multi-vehicle (MV) crashes. This study positions itself at exploring the differences from spatiotemporal, road-environment, driver-vehicle, and collision characteristics. A model comparison as well as the elasticities for the optimal model (partial proportional odds model) is implemented based on 18,083 SV crashes and 22,162 MV crashes in China. The results evidenced the great differences that time, road, speed, lighting, and weather are found to have a positive correlation with only SV crash injury severity, yet negatively related with only MV crash injury severity. Area, location, and angle are significant only for SV crashes, while day, interference, and wind are significant only for MV crashes. The findings revealed that gender, age, collision, location, and time are more influencing factors in SV crashes, while collision, age, gender, vehicle, and wind have more contributions to MV crashes. The findings could provide an insightful reference for prioritizing effective countermeasures to mitigate traffic crash losses.