Professional Documents
Culture Documents
net/publication/335364048
CITATIONS READS
0 62
2 authors, including:
Csaba Csiszar
Budapest University of Technology and Economics
100 PUBLICATIONS 326 CITATIONS
SEE PROFILE
Some of the authors of this publication are also working on these related projects:
Method for selection of appropriate location of charging stations for electric vehicles View project
All content following this page was uploaded by Csaba Csiszar on 27 August 2019.
S. Anis and Cs. Csiszár are with the Department of Transport Technology (BME), Budapest, Hungary (e-mails: summairuetian66@gmail.com,
and Economics (KUKG), Faculty of Transportation Engineering and Vehicle csiszar.csaba@mail.bme.hu).
Engineering (KJK), Budapest University of Technology and Economics The publication of the work reported herein has been supported by ETDB at
BME.
Questionnaire Surveys
concluded that most of the respondents feel more comfortable PROCESS
with using the semi-automated AV. AHP Method
It is found that these conflicts should be identified and Data Sets
prioritized, in order to increase the safety for non-motorized Prioritization of Conflict
Factors OUTPUT
road users. This prioritization can be done by taking opinions Stakeholders Input
Suggested Mitigation
from the stakeholders working in this field. The best tool to Measures NEXT STEP
accomplish this task is the questionnaire survey. The survey TWO WAY CONNECTION
results highlight the major improvement opportunities to be INPUT FROM PREVIOUS STEP
carried out in the road infrastructure. The results also highlight Phase III (Data Analysis)
important aspects of pedestrian behaviors that needs to be AHP Method
Normalization of Datasets
studied, in terms of their interactions with AVs. In this way, Calculation of Weight Factors
safe and efficient operation of AVs can be achieved. Determination of Qualities
Estimation of Choice Probabilities
Vehicle
where:
Industry
Industry N’ij are values of entries in the table after normalization,
Transport Nij are the original values of the entries,
Operator ΣNj is the sum of all the entries in a column.
Weightage factors are determined for sub-factors, which are
Fig. 2. Identified key stakeholders for questionnaire survey represented by a row in each table, using (2):
B. Data Analysis (Phase III):
∑
Several techniques can be applied to analyze data, including gi= (2)
mutual correspondences, regression analyses, AHP methods,
etc. Among these different techniques, AHP methods are where:
effective for use when a rating of factors according to their gi is the weight factor for sub factors,
comparison with other factors is needed. Using the AHP ∑N’i is the sum of the normalized values in each row,
method, the final goal is the calculation of choice probabilities n is the total number of entries in each row.
of factors based on the sub-factors. For this purpose, different C. Calculation of Qualities:
steps (e.g. normalization, weight factor calculations, etc.) have
been carried out as explained in section V. The term “quality” is a standard term used in an AHP. This
term represents the effect of a sub-factor on the particular
V. RESULTS & DISCUSSION factor, according to the weightage. Quality is further used for
the calculation of choice probability of factors. The calculated
A. Application of AHP Method: quality values are presented in Table I and Table II. The quality
Two tables are filled with factors from the stakeholders. In values are determined as follows (3):
the first table, each factor is rated from 1-5 according to the
importance. This table gives linear ratings. In the second table, Qi = ∑ g ∗ r (3)
each factor is compared with the other one and the score from
1-5 is given as relative importance. This table gives relative where:
ratings [10]. The rating scale of 1-5 represents the following Qi is the quality value for each factor,
values: gi is the weight factor for each sub-factor,
1. not important, ri is the linear rating of each sub-factor (this represents
2. less important, the scoring of each sub-factor based on its own
3. equal weightage, uniqueness without the comparison with other sub-
4. important, factors).
5. very important. In Table I, sub-factors are given by their codes (see appendix
I). Linear rating values are indicated by ri, while relative rating
values are indicated by gi.
TABLE I
QUALITIES OF PEDESTRIAN SIDE FACTORS
Illegal Crossings Collision Avoidance Decisions Drunk/ Tired Pedestrians Poor Road Infrastructure
Factors Factors Factors Factors
ri gi ri gi ri gi ri gi
Codes Codes Codes Codes
P.1.1 4.1 0.64 P.2.1 3.7 0.88 P.3.1 4.3 0.28 P.4.1 4.1 0.13
P.1.2 3.2 0.44 P.2.2 4.3 1.14 P.3.2 3.8 0.23 P.4.2 3.7 0.14
P.1.3 4.3 1.04 P.2.3 3.5 0.99 P.3.3 3.1 0.22 P.4.3 3.2 0.14
P.1.4 4.1 0.80 P.2.4 4 0.86 P.3.4 3.5 0.27 P.4.4 3.6 0.14
P.1.5 2.7 0.38 P.4.5 3.8 0.13
P.1.6 3.3 0.43 P.4.6 3.5 0.10
P.4.7 4 0.13
P.4.8 3.4 0.10
Qi 3.72 3.87 3.70 3.67
TABLE II Conflict Factors
QUALITIES OF AV SIDE FACTORS
Identification of Collision Avoidance Barriers for the Analysis of factors on Analysis of factors on AV
Pedestrians Decisions Operation Pedestrian Side Side
Factors Factors Factors
ri gi ri gi ri gi
Codes Codes Codes
A.1.1 3.4 0.75 A.2.1 4 0.25 A.3.1 4.2 0.18 Calculation of Choice Probabilities
A.1.2 3.7 1.04 A.2.2 4.1 0.27 A.3.2 3.5 0.17
A.1.3 3.1 0.88 A.2.3 4.3 0.26 A.3.3 3.2 0.15 Factors Sub-Factor
A.1.4 2.9 0.62 A.2.4 4 0.22 A.3.4 3.1 0.19
A.3.5 4.6 0.14
NO Are Values of Choice Are Values of weights NO
A.3.6 4.7 0.18
Probability >10% >10%
Qi 3.29 4.10 3.87
YES YES
D. Calculation of Choice Probabilities:
Severity Levels:
In order to determine the relative importance of the factors, >10% - < 30%----------Less Important
Particular Sub-Factor to be
Ignored, >30% - < 50%-----------------Important
the choice probability values are calculated. It shows the Not Important >50% - < 70%----------Very Important
Ignored,
Not Important
percentage by which one factor is more important than others. >70% - < 90%------- Actions Required
>90%---Top Priority, Immediate Actions
It can be calculated by (4): LEGEND
Decision Making
Mitigation
Measures
Pi = ∑ (4) Process
Poor Road Infrastructure 3.67 12.94 Poor Road Infrastructure 12.94 Important
AV Side
Identification of Pedestrians by AVs 8.83 Ignored
Identification of Pedestrians 3.29 8.83
Side
P.1.3 24.19 transit stops for the prioritized factors. It can be decided which factors are to
P.1.4 19.54 • Installation of variable message signs be considered for mitigation in the long term and short term
• Coding of pedestrian behavioral data
P.1.5 13.97 into the AVs
based on the severity levels.
• Improvement of the ability of AVs to As a result of questionnaire survey the following key findings
P.1.6 13.08 locate (and apply breaks automatically are identified:
upon seeing) a pedestrian • The most important factors which cause conflicts
• Teaching pedestrians for improvement
Decisions by Pedestrians
A.2. Collision Avoidance P.2. Collision Avoidance
P.2.1 23.83 of road crossing behavior between pedestrians and AVs are the decision making
• Increasing awareness regarding road abilities of AVs and road crossing behavior of the
P.2.2 26.40 safety pedestrians.
• Observation of pedestrian movement
trajectories over time and programming
• It has been identified that current road structure is
P.2.3 28.30
them into the AVs suitable for the operation of AVs; only a few
P.2.4 21.47 • Provision of separated pedestrian modifications are required.
facilities (underpass/overpass)
• Applications of suggested mitigation measures in the
• Separation of pedestrian and normal
A.2.1 25.30 traffic during peak periods physical form can guarantee the successful operation of
Decisions by AVs
A.3.2 16.83 conditions even if the traffic signals are not 2017. DOI: 10.1155/2017/1723728
A.3.3 14.94 working [3] D.F. Llorca, V. Milanes, I.P. Alonso, M. Gavilan, I.G.
AVs
A.3.4 18.59 • Improvement of the hazard management Daza, J. Perez, and M.Á. Sotelo, “Autonomous Pedestrian
system for the AVs
A.3.5 14.21 Collision Avoidance sing a Fuzzy Steering Controller,”
A.3.6 17.57 IEEE Transactions on Intelligent Transportation Systems,
• Automatic repair system for AVs to vol. 12, no. 2, pp. 390-401. 2011. DOI:
determine and restore faults automatically.
• Ability of the AVs to send error reports
10.1109/TITS.2010.2091272
to the developers in case of failure [4] J. Levinson, J. Askelnad, J. Becnker, J. Dolson, D. Held,
A.1. Identification of
• Installation of extra cameras for the AVs S. Kammel, J.Z. Kolter, D. Langer, O. Pink, V. Pratt, M.
Pedestrians by AVs