自动驾驶行为决策算法综述

Development Review of Behavior Decision Algorithms in Autonomous Driving

行为决策作为当前主流自动驾驶框架中的重要组成部分,承担了环境感知和运动规划之间的桥梁作用,也是实现车辆智能化的关键所在。本文将行为决策算法依照维度分为宏观和微观两个层面,并进一步根据决策对象、输出表征类型和方法进行分类,介绍了不同分类下的行为决策算法的主流实践和方法,分析其特点,并以此为基础对该领域未来的发展重点提出了展望。

Behavior decision is becoming an important component in the common autonomous driving(AV)architecture,which acts as a bridge between environment modelling and motion planning.It plays a key role in the intelligence of the AV We divide the behavior decision into macroscopic and microscopic parts.Furthermore,the behavior decision is classified into more subcategories according to the decision object,output type and method.Popular practices and algorithms are introduced and then analyzed.Then we proposed an outlook of the development of behavior decision in the future.