Abstract: To improve the accuracy of adaptive cruise control (ACC) target selection in scenarios such as turning, lane missing, and ultra wide lanes, a target selection algorithm based on road models and scene judgment is proposed. Establishing optimization problems through the Ceres optimization library, fully considering factors such as self vehicle historical trajectory, target vehicle historical trajectory, self vehicle steering wheel angle, vehicle speed, lane line, road edge, etc., and assigning different weights to generate a self vehicle pre driving reference line that can be used as the basis for ACC target screening. For different scenarios, using the self vehicle pre driving reference line as the benchmark to establish risk zones, judging the relative position relationship between each target location point and the danger zone through geometric relationships, determining it as a following target, entering and exiting the target, responding as soon as possible, and improving driving experience and safety.

Key words: reference line; target screening; Ceres optimization; hazardous areas; driving scene

摘要： 为提高转弯、车道线缺失、超宽车道等场景下自适应巡航控制（ACC）跟车目标筛选 的准确性，提出一种基于道路模型与场景判断的目标筛选算法。通过 Ceres 优化库建立优化 问题，充分考虑自车历史轨迹、目标车历史轨迹、自车方向盘转角、车速、车道线、路沿等 因素，分别配与不同权重，生成一条可用于 ACC 目标筛选基础的自车预行驶参考线。针对不 同场景，以自车预行驶参考线为基准建立风险区，通过几何关系判断各目标位置点与危险区 的相对位置关系，判断为跟车目标、切入、切出目标，尽早响应，提高驾乘体验及安全性。

关键词: 参考线；目标筛选；Ceres 优化；危险区；驾驶场景