Threat Assessment Techniques in Intelligent Vehicles: A Comparative Survey
Yang Li, Keqiang Li, Yang Zheng, Morys Bernhard, Shuyue Pan, Jianqiang Wang
Abstract
Threat assessment evaluates the situational criticality and helps guarantee the driving safety of intelligent vehicles. Many critical metrics have been proposed for threat assessment, and it is essential to choose a suitable critical metric to address specific driving tasks under different driving conditions. This study presents a comparative overview of the state-of-the-art critical metrics based on real-world driving data validations. The critical metrics are categorized into five groups, i.e., time-based metrics, kinematics-based metrics, statistics-based metrics, potential field-based metrics, and unexpected driving behavior-based metrics. Benefits and limitations of the critical metrics are compared and validated based on the real-world data from naturalistic driving study, field operational test, and numerical simulations. Three typical driving scenarios, i.e., car-following, cut-in, and pedestrian-crossing, are extracted to evaluate the effectiveness of the critical metrics. Challenges and future researches of the critical metrics are also highlighted. This comprehensive and comparative overview is expected to assist with a system-level design for threat assessment of intelligent vehicles.