PSYCH OpenIR  > 社会与工程心理学研究室
数据挖掘在汽车驾驶安全及消防听觉导航模式研究中的应用
Alternative TitleThe Application of Data Mining in Driving Safety and Audio Navigatio Mode in Firefighting
于路
2011-10
Publication Place北京
Contribution Rank1
Abstract

随着计算机技术的普及和超大容量数据存储技术的发展,人们在各行各业的应用和研究中,积累的数据呈爆炸式增长。为了从这些海量的数据中发现有价值的知识及规律,人们综合统计学、数据库、机器学习等领域的方法,提出数据挖掘技术来解决这一难题。

工程心理学研究中常常涉及大量的行为和生理数据,充分挖掘这些数据包含的信息,进一步深入分析,是研究工作的迫切需求。本文基于数据挖掘技术,对驾驶分心、驾驶技能模式和消防员听觉导航系统显示界面进行了深入的研究,具体内容如下:

一、运用数据挖掘中的模式识别方法进行了驾驶员分心状态和驾驶技能模式的分析和研究。驾驶分心是引发交通事故的重要隐患,本文分别提取驾驶员驾驶绩效数据和生理数据作为特征指标,运用隐马尔科夫模型与boosting方法相结合的分类算法和支持向量机方法分别利用驾驶绩效指标和生理指标对驾驶员的分心状态进行了检测,结果表明:运用驾驶员驾驶绩效指标和生理指标进行驾驶分心检测是可行的。驾驶员驾驶技能模式的识别对于车载驾驶辅助系统的研究有着重要意义,本文利用驾驶员驾驶操作数据,结合隐马尔科夫模型对驾驶员驾驶技能进行了识别,平均正确率达到了80%以上。

二、运用数据挖掘中时间序列的统计分析方法研究了消防员火灾现场听觉一导航系统的信号发送方式。火灾现场尤其是建筑物内的火灾现场充满烟雾,听觉导航系统会给消防员灭火和搜救被困人员提供重要帮助。本文在计算机模拟环境下考察了消防员在听觉导航系统指引下的行为,对消防员的完成灭火和搜救任务的过程进行了精细的编码,重点对消防员实际行走路径的平稳性以及消防员在不同导航阶段的行为进行了精确描述的和量化分析,充分挖掘出隐藏在消防员完成任务过程中的路径中的内在信息。其中对不同的听觉导航模式:纯语音导航、纯3D声音导航、转弯阶段使用语音,直行阶段使用3D声音、转弯阶段使用3D声音、直行阶段使用语音,共四种一导航模式进行了对比,得出结论:在转弯阶段和直行阶段使用不同的导航模式比在两个阶段使用相同的模式导航效率高;转弯阶段使用语音导航,直行阶段使用虚拟3D声音的混合导航模式效率最高。

Other Abstract

The development of computer and super capacity data storage technology lead to the explosive growth of data accumulation in various production and research field. Data mining technology, which synthesized statistics, data base and machine learning, ctc., was employed to seek out the useful information from the large mass of data. Engineering psychology research is often involved in great amount of behavior and physiological data. Finding the useful information implies in the data is the needs of the researches. In this papcr9 data mining technology was employed to analyze the experimental data in driving and the display mock of navigation systems for firefighters.

Some useful results were derived.

The work in this paper includes two pants:

1 .The research of driving distraction and driving skill in which the recognition method was used. Driving distraction is an essential cause lead to traffic accident. In this paper, the discriminating feature of drivers' distraction state which included driving performance and physiological signal were extracted. Then, Hidden Markov Model combined with boosting algorithm was used to detect driving distraction by the performance feature and physiological data respectively. The results indicated that it was feasible to detect driving distraction by performance data or physiological data. On the other hand, driving skill mode was important for driving assistant system. In this paper, Hidden Markov Model was employed to differentiate the experienced drivers from novice drivers. The accuracy of the recognition was more than 80%.

The research of the signal transition mode in navigation systems for firefighters. The statistical analysis method of time series was used in the research. Audio navigation system is essential for firefighters in an afire building full of fog. In this paper, an indoor route guidance task in a virtual environment realized by desktop virtual reality technique to investigate the efficacy of audio spatial display modes in indoor navigation systems was reported. Four audio display modes including speech, 3D-audio and two mixed modes which combined 3D-audio display with speech and non-speech signal according to their roles in the navigation as guiding straight walking or turning were tested and compared. All of the evaluation metrics and subjective ratings converged to the fact that mixed display modes had clear advantages than single display modes.

Keyword驾驶分心 驾驶技能模式 听觉导航系统 数据挖掘 隐马尔科夫模型 Boosting 支持向量机
Pages43
Language中文
Document Type科技报告
Identifierhttp://ir.psych.ac.cn/handle/311026/29237
Collection社会与工程心理学研究室
Affiliation中国科学院心理研究所
Recommended Citation
GB/T 7714
于路. 数据挖掘在汽车驾驶安全及消防听觉导航模式研究中的应用[R]. 北京,2011.
Files in This Item:
File Name/Size DocType Version Access License
于路-博士后研究工作报告.pdf(3181KB)科技报告 限制开放CC BY-NC-SAView Application Full Text
Related Services
Recommend this item
Bookmark
Usage statistics
Export to Endnote
Google Scholar
Similar articles in Google Scholar
[于路]'s Articles
Baidu academic
Similar articles in Baidu academic
[于路]'s Articles
Bing Scholar
Similar articles in Bing Scholar
[于路]'s Articles
Terms of Use
No data!
Social Bookmark/Share
File name: 于路-博士后研究工作报告.pdf
Format: Adobe PDF
All comments (0)
No comment.
 

Items in the repository are protected by copyright, with all rights reserved, unless otherwise indicated.