Safety represents the fundamental bottleneck in the development of autonomous driving.The majority of current research focuses on driving safety in regular scenarios. However, certain actions by other vehicles may pose significant risks to the ego vehicle, rustling in emergency scenarios. In this paper, we propose an emergency escape strategy for autonomous vehicles, encompassing risk assessment, escape trajectory generation, and Model Predictive Control (MPC) based control schemes. In risk assessment module, we introduce the Fusion Time-toCollision (FTTC) concept for risk quantification. The proposed approach not only considers the maximum escape potential under extreme circumstances but also incorporates vehicle dynamics constraints during the escape process to ensure that risks such as rollover are avoided due to excessively aggressive escape trajectories.The performance of our proposed algorithm is measured in two simulation scenarios.