自动驾驶汽车-行人交互研究综述

胡宏宇 ¹ ，刁小桔 ² ，高菲 ¹ ，高振海 ^1，2

1.吉林大学，汽车仿真与控制国家重点实验室

2.吉林大学，汽车工程学院

【摘要】 自动驾驶汽车已开始在部分开放道路进行测试，其与行人等其他交通参与者共享混行道路，面向自动驾驶汽车的车外人机交互技术亟需开展深入研究，以便行人快捷、高效地理解自动驾驶汽车的行驶意图，确保混合交通场景通行安全并提高通行效率。本文首先阐述了自动驾驶汽车与行人交互的重要意义，并从行人检测与跟踪、行人意图识别及行为预测、自动驾驶汽车的决策3个方面介绍了目前自动驾驶汽车的车外人机交互前期支撑技术研究概况，着眼于自动驾驶汽车行驶意图的表达，对交互需求和车外人机交互界面的设计原则及质量评估进行了梳理，最后提出了车外人机交互面临的挑战及未来的发展方向。

【关键词】 自动驾驶汽车，人机交互，行人意图识别，行为预测

Review of Interaction between Autonomous Vehicles and Pedestrians

Hu Hongyu ¹ ，Diao Xiaoju ² ，Gao Fei ¹ ，Gao Zhenhai ^1，2

1. State Key Laboratory of Automobile Simulation and Control ， Jilin University

2. School of Automotive Engineering ， Jilin University

Abstract: Autonomous Vehicles（AVs）have begun to be tested on some open roads，and are mixed with other road participants such as pedestrians to share the road.There is an urgent need for in-depth research on the interaction technology between AVs and pedestrians，so that human can quickly and efficiently understand the driving intention of AVs，to ensure the safety and efficiency of mixed traffic.This article first introduces the significance of interaction between AVs and pedestrians.In addition，the current supporting technology hotspots of external human-machine interaction of AVs are introduced from three aspects: pedestrian detection and tracking，pedestrian intention recognition and behavior prediction，and AVs' decision-making.Based on the expression of AVs' driving intention，the interaction requirements and the design principles and quality evaluation of external Human-Machine Interface（eHMI）are sorted out.At last，the article puts forward eHMI challenges and future development directions.

Key words: autonomous vehicles（AVs），human-machine interaction，pedestrian intention recognition，behavior prediction wlUm/NyvgP4kvIXtbuqGRmmlEfPWKjzdjYzBvtwV28OD9D3l9ZDocE/SoG9dxJJF