• CN:11-2187/TH
  • ISSN:0577-6686

机械工程学报 ›› 2023, Vol. 59 ›› Issue (20): 281-303.doi: 10.3901/JME.2023.20.281

• 运载工程 • 上一篇    下一篇

扫码分享

智能汽车环境感知方法综述

彭湃, 耿可可, 王子威, 柳智超, 殷国栋   

  1. 东南大学机械工程学院 南京 211189
  • 收稿日期:2023-06-13 修回日期:2023-08-30 出版日期:2023-10-20 发布日期:2023-12-08
  • 通讯作者: 殷国栋(通信作者),男,1976年出生,博士,教授,博士研究生导师。主要研究方向为先进电动汽车、车辆动力学与控制、智能汽车和车辆主动安全控制。E-mail:ygd@seu.edu.cn
  • 作者简介:彭湃,男,1993年出生,博士研究生。主要研究方向为智能汽车多模态融合感知。E-mail:pengpai@seu.edu.cn
  • 基金资助:
    国家自然科学基金资助项目(51975118, 52025121, 52272414)。

Review on Environmental Perception Methods of Autonomous Vehicles

PENG Pai, GENG Keke, WANG Ziwei, LIU Zhichao, YIN Guodong   

  1. School of Mechanical Engineering, Southeast University, Nanjing 211189
  • Received:2023-06-13 Revised:2023-08-30 Online:2023-10-20 Published:2023-12-08

摘要: 智能汽车是全球汽车产业的未来发展方向,是推动我国自主汽车产业高质量发展的应有之义。智能汽车依靠人工智能、泛在传感等先进技术的赋能,实现对驾驶人认知感知、决策规划及控制执行的增强或替代。对道路环境的实时、精准和鲁棒的感知是汽车智能化的基石,近十年间智能汽车感知领域的巨大飞跃多是由深度学习技术推动的。针对近年智能汽车环境感知技术的发展现状,首先梳理智能汽车环境感知系统软硬件架构,对感知、计算设备以及算法部署平台进行总体概述;其次,围绕目标检测与语义分割、多目标跟踪、目标意图识别与轨迹预测、环境建图四方面关键任务,总结近年具有里程碑意义的研究方法与技术方案;最后,分析当前智能汽车环境感知技术所面临的问题和挑战,并对未来发展趋势与关键技术进行展望。

关键词: 智能汽车, 环境感知, 目标检测与语义分割, 跟踪和预测, 环境建图

Abstract: Autonomous vehicles are the future development direction of the global automotive industry and are essential for promoting the high-quality development of China's independent automotive industry. Autonomous vehicles rely on advanced technologies such as artificial intelligence and ubiquitous sensing to enhance or replace the driver's cognitive perception, decision-making planning, and control execution. Real time, accurate, and robust perception of the road environment is the cornerstone of automotive intelligence, and the huge leap in the field of autonomous vehicles perception in the past decade has been mostly driven by deep learning technology. A review of the development of autonomous vehicles environmental awareness technology in recent years is provided. Firstly, it summarizes the software and hardware architecture of autonomous vehicles environmental awareness systems, and provides an overall overview of perception, computing devices, and algorithm deployment platforms; Secondly, milestone research methods and technical solutions in recent years were summarized around four key tasks: object detection and semantic segmentation, multi objects tracking, object intention recognition and trajectory prediction, and environmental mapping; Finally, the problems and challenges faced by current autonomous vehicles environmental perception system were analyzed, and future development trends and key technologies were prospected.

Key words: autonomous vehicles, environmental perception, object detection and semantic segmentation, tracking and prediction, environmental mapping

中图分类号: