Integrated Longitudinal and Lateral Acceleration-Based Trajectory Planning Algorithm for Emergency Collision Avoidance Maneuver in Autonomous Vehicle

초록

While autonomous driving technology has significantly enhanced the safety and convenience of driving in the automotive industry, the field of collision avoidance technology, especially in emergency situations, still requires advanced development. Statistical analysis of accidents involving level 2 autonomous driving technology reveals limitations in the current collision avoidance systems. The existing systems, such as Autonomous Emergency Braking (AEB) and Autonomous Emergency Steering (AES) used in level 2 autonomous vehicles, primarily focus on the forward vehicle, indicating a constraint in addressing dynamic situations involving surrounding vehicles. To address these limitations, there is a need for emergency collision avoidance technology that considers the surrounding environment and the driving conditions of the ego vehicle. This paper presents an acceleration-based emergency collision-avoidance method for autonomous vehicles in multi-vehicle scenarios. The proposed acceleration-based trajectory planning (ATP) method evaluates evasive maneuver candidates on an ego-centric local risk map and generates coupled longitudinal and lateral accelerations for vehicle motion. Unlike geometry-first planning followed by a separate speed-planning stage, the ATP framework formulates emergency collision avoidance as an integrated maneuver-generation problem under a risk-triggered condition. The method was validated in representative structured multi-vehicle emergency scenarios and compared with conventional emergency collision-avoidance systems, including autonomous emergency braking (AEB) and autonomous emergency steering (AES). The results show that the ATP generates safer evasive maneuvers than conventional emergency collision-avoidance systems by considering surrounding vehicles in multiple directions.

키워드