电子学报 ›› 2016, Vol. 44 ›› Issue (2): 365-373.DOI: 10.3969/j.issn.0372-2112.2016.02.018

• 学术论文 • 上一篇    下一篇

基于云模型的智能驾驶车辆变粒度测评研究

高洪波1,2, 张新钰2, 张天雷3, 刘玉超4, 李德毅1,3,4   

  1. 1. 北京航空航天大学软件开发环境国家重点实验室, 北京 100191;
    2. 清华大学信息技术中心, 北京 100083;
    3. 清华大学计算机学院, 北京 100083;
    4. 电子系统工程第61研究所, 北京 100039
  • 收稿日期:2014-12-17 修回日期:2015-10-28 出版日期:2016-02-25 发布日期:2016-02-25
  • 通讯作者: 高洪波
  • 作者简介:张新钰 男,1979年10月出生,甘肃会宁人,现为中国人工智能学会副秘书长,剑桥大学访问学者.主要研究方向为智能机器人与网络教育E-mail:xyzhang@mail.tsinghua.edu.cn
  • 基金资助:

    国家863计划(No.2015AA015401);国家自然科学基金重大研究计划专项(No.90920305);国家自然科学基金(No.61305055,No.61035004,No.61273213,No.61300006,No.61203366);中国工程院项目(No.2015-XY-42)

Research of Intelligent Vehicle Variable Granularity Evaluation Based on Cloud Model

GAO Hong-bo1,2, ZHANG Xin-yu2, ZHANG Tian-lei3, LIU Yu-chao4, LI De-yi1,3,4   

  1. 1. State Key Laboratory of Software Development Environment, Beihang University, Beijing 100191, China;
    2. Information Technology Center, Tsinghua University, Beijing 100083, China;
    3. Department of Computer Science and Technology, Tsinghua University, Beijing 100083, China;
    4. The 61th Institute of Electronic System Engineering, Beijing 100039, China
  • Received:2014-12-17 Revised:2015-10-28 Online:2016-02-25 Published:2016-02-25

摘要:

如何通过直观、简单和有效的测评方法,建立一个定性与定量的不确定性测评转换模型,实现智能驾驶车辆的测评研究,已成为智能驾驶车辆研究领域急需面对和解决的一个基本问题,也是一个难题.为了解决这个问题,本文提出一种基于云模型与变粒度的测评方法.首先,提出一套4S变粒度测评体系与三级智商变粒度测评体系;其次,通过云模型将智能驾驶车辆的定性测评转化为直观、形象的定量测评,利用云模型的期望、熵与超熵对智能驾驶车辆进行定量评价,从而建立一个由定性到定量的测评转换模型.为解释这个模型的实施,2013年未来挑战赛的实例被采用.实验的分析表明,在具有定性与定量的测评中,利用云模型的期望、熵与超熵作为定性评价的定量评价的依据,并且,基于云模型与变粒度的智能驾驶车辆的测评能够有效地解决多目标粒度的测评与描述性的定性评价的定量评价,从而解决了一类智能驾驶车辆的测评研究问题.

关键词: 云模型, 变粒度, 智能驾驶车辆, 测评研究

Abstract:

A direct, easy-to-operate and effective evaluation method for building a qualitative and quantitative uncertainty evaluation transformation model is urgently needed for intelligent vehicle evaluation research.It has been a challenging problem facing the researchers.Based on cloud model and variable granularity, this paper describes a measurement method.First, we proposed a set of 4S variable granularity measurement system and the three-level IQ variable granularity evaluation system.Then, we transformed the qualitative assessment over the intelligent driving vehicle into a direct image quantitative evaluation via using the Expectation, the Entropy and the Hyper Entropy of cloud model, thus building an uncertainty transformation model from qualitative to quantitative.The Future Challenge 2013 was used as a case to illustrate the model.The analysis shows that in the qualitative and quantitative evaluation, the Expectation, Entropy and Hyper Entropy based on cloud model is the basis of qualitative-evaluation of quantitative evaluation.Moreover, the intelligent vehicle evaluation based on cloud model and variable granularity can efficiently solve the multi-objective particle size measurement of evaluation and descriptive qualitative evaluation of quantitative evaluation, thereby providing solutions to the evaluation of a class of intelligent vehicle research.

Key words: cloud model, variable granularity, intelligent vehicle, evaluation research

中图分类号: