Collision avoidance is a key issue for high-speed vehicles in emergency situations, particularly as it pertains to surrounding vehicles. Accordingly, a hierarchical framework for human-driven or autonomous vehicles is proposed that ensures the safe operation of vehicles in emergency driving scenarios while considering surrounding vehicles. The developed emergency collision-avoidance system consists of an estimator module, a prediction module, a manoeuvre decision-making module, and a manoeuvre control module. The core module for manoeuvre decision making uses finite state machine (FSM) technology to determine the appropriate manoeuvre for avoiding collisions. In particular, a collision risk model is developed in this module, taking into account risk related to surrounding vehicles in the overlapping region, road adhesion, and stabilization performance of vehicle. In the manoeuvre control module, accounting for safe gaps to surrounding vehicles in an adjacent lane, the longitudinal manoeuvre is generated by a model predictive-control algorithm. A yaw stability controller is designed using lateral tire force estimate-based sliding mode control, accounting for both the collision and vehicle stabilization. To mediate the demands of yaw stabilization and collision risk, particularly for reducing the loss of velocity, an integrated controller with propulsion is developed. Modeling vehicles driving under different road conditions with various road adhesion and surrounding-vehicle settings, the results of experiments in a hardware-in-the-loop (HIL) system have successfully demonstrated the effectiveness of the proposed collision avoidance system.