(A) study of stochastic estimation method for driver acceptance evaluation of autonomous driving systems자율주행시스템의 운전자 수용성 평가를 위한 확률적 추정방법 연구

Cited 0 time in webofscience Cited 0 time in scopus
  • Hit : 463
  • Download : 0
Recently, with increased interest in high levels of autonomous driving systems such as automatic lane change system, the need for reliable assessment methods of driver acceptance has arisen. Because the acceptance depends on the individual, the driver acceptance assessment can only be based on individuals’ personal attitudes such expectations and experiences, etc. Accordingly, subjective evaluation methods have mostly been utilized to assess the acceptance of newly developed advanced driver assistance systems. In this dissertation, an objective evaluation methodology of the driver acceptance for autonomous driving system is proposed based on objective measurable parameters. In order to achieve the objective, a specific experimental program is designed and conducted for a limited condition that the target system is not available, currently. Throughout the experiment, a massive database is constructed with 19 selected drivers; the database consists of not only various measurable parameters on control commands, vehicle behaviors, and vehicle relations, but also subjective driver acceptance. The relations between vehicles are considered as the most influential environmental factor on the driver acceptance. Thus, the relative vehicle relations considering road characteristics with respect to road characteristics are accurately identified by suggesting specific method. A specific algorithm for identification of lane change sections is also proposed for dealing with thousands of the driving events. Based on the prepared data sets, in order to interpret the driver acceptance, objective parameter sets are derived by two different methods: statistical significance test, and acceptance sensitivity analysis. Then, driver acceptance evaluation modeling method is suggested based on stochastic estimation of the driver acceptance as an objective evaluation method of the driver acceptance with the derived objective parameters. The previously derived parameter sets are regarded as modeling parameters for modeling of the driver acceptance. The developed models are experimentally verified by applying to other specific data sets having different driving characteristics. And, the evaluation and modeling method is also validated on individuals’ assessment characteristics by modeling a personalized acceptance evaluation model with specific set of data, evaluated by only one evaluator, in consideration of the recently increased interests in personalized autonomous driving systems as a solution of maximum acceptance of autonomous driving systems. In most of modeling and verification results, the driver acceptance evaluation models using the acceptance sensitivity analysis not only estimates well, but also is more robust to both sample sizes of the data sets for modeling and certain characteristics of the data sets. Results of this study can contribute to the development process of autonomous driving system in several features: the design targets can be clearly defined by the proposed parameter selection method with respect to the driver acceptance, the driver acceptance can be objectively assessed with objective parameter values, and modeling methodology of this proposed driver acceptance evaluation model is applicable to the development of personalized autonomous driving system.
Advisors
Choi, Seibumresearcher최세범researcher
Description
한국과학기술원 :기계공학과,
Publisher
한국과학기술원
Issue Date
2018
Identifier
325007
Language
eng
Description

학위논문(박사) - 한국과학기술원 : 기계공학과, 2018.2,[ix, 152 p. :]

Keywords

driver acceptance▼aobjective evaluation▼adriver acceptance evaluation model▼astochastic estimation▼aautonomous driving systems; 운전자수용성▼a객관평가▼a운전자수용성 평가모델▼a확률적 추정▼a자율주행 시스템

URI
http://hdl.handle.net/10203/264570
Link
http://library.kaist.ac.kr/search/detail/view.do?bibCtrlNo=734245&flag=dissertation
Appears in Collection
ME-Theses_Ph.D.(박사논문)
Files in This Item
There are no files associated with this item.

qr_code

  • mendeley

    citeulike


rss_1.0 rss_2.0 atom_1.0