MuLSA: Multi-Layered Scenario Analysis for

an Advanced Driver Assistance System

Takako Nakatani

and Keita Sato

Graduate School of Business Sciences, University of Tsukuba, 3-29-1, Otsukba, Bunkyo-ku, Tokyo, 112-0012, Japan

DENSO Corporation, 1-1, Shouwacho, Kariya-shi, Aichi, 448-8661, Japan

Keywords:

Requirements Engineering, Customer Journey Map, Scenario Analysis, Cruise Control System, Advanced

Driver Assistance System.

Abstract:

Safe driving is mandatory for an advanced driver assistance system (ADAS). We consider that the adequacy

and safety of the driver assistance services can be monitored by observing drivers’ positive and negative

emotions, since, if they ﬁnd a hazardous situation, he/she has a negative emotion, ”surprised” or ”dangerous.”

If they feel to be assisted by the system, they will have a positive emotion. In order to analyze requirements

for the ADAS, we propose a multi-layered scenario analysis (MuLSA). MuLSA is developed by integrating a

customer journey map and a service blueprint with the context of a scenario. A customer journey map consists

of a scenario; as the customer’s experiences, as well as a customer’s emotions toward the services. The service

blueprint represents a customer’s journey, as well as a mechanism of the services. Thus, MuLSA consists of

a driver’s journey, his/her emotions, the mechanism of services, as well as the context of the service. In order

to prioritize requirements for the safety of a future ADAS, we have observed driver emotions with regard

to hazardous scenarios with MuLSA. This paper shows the results of the observation, and we discuss the

effectiveness of MuLSA.

1 INTRODUCTION

We develop services and improve them to com-

pete in the market. According to the SQuaRE

(Software product Quality Requirements and Evalu-

ation) (ISO/IEC 25000:2005, 2005), customer satis-

faction is evaluated as the satisfaction in the usability

of services. We need a method to analyze the satis-

faction of customers of the current system in order

to prioritize requirements of the future system. In this

paper, we introduce a method to elicit requirements of

an Advanced Driver Assistance System (ADAS) base

on the analysis of the current Adaptive Cruise Control

system (ACC).

The Cruise Control system (CC) of a car is one of

the driver assistance systems (DAS). It regulates the

speed of a car. In general, the structure of such ser-

vices consists of two layers: one is a service provider,

while the other is a service receiver. CC is a service

provider. It provides services, i.e. start, termination,

and acceleration, to the driver who is the service re-

ceiver. The service of CC is provided by a request

from the driver. If we can observe the ups and downs

of emotions of a driver when utilizing the CC, we can

evaluate the customer’s satisfaction of the CC. How-

ever, the structure of the latest driving assist system

is not so simple. The ACC regulates the speed of the

car, while also regulating the distance from the prece-

dent car. The structure of the service of ACC con-

sists of the precedent car, as well as the driver and the

ACC. The lane departure warning system is another

example of DAS. It helps the car navigate the trafﬁc

lane. In this case, the structure of the service analy-

sis has to take into account the trafﬁc lane. Some of

these systems stop their services in heavy rain, since

they cannot monitor the precedent car or the trafﬁc

lane under such a bad weather conditions. Thus, the

weather must also to be a consideration within the

service structure. ADAS provides more complex ser-

vices than DAS and the lane departure warning sys-

tem. It is able to monitor peripheral cars, load condi-

tions, trafﬁc lights, trafﬁc signs, etc. and, make deci-

sions in order to ensure the safety of the driver’s jour-

ney. The purpose of this paper is to develop a scenario

analysis method for the DAS in order to prioritize re-

quirements of the ADAS. The method will help us

evaluate the effects of driver emotion with regard to

current services, and the quality of these services. We

Nakatani T. and Sato K..

MuLSA: Multi-Layered Scenario Analysis for an Advanced Driver Assistance System.

DOI: 10.5220/0004449500830091

In Proceedings of the 8th International Joint Conference on Software Technologies (ICSOFT-EA-2013), pages 83-91

ISBN: 978-989-8565-68-6

 2013 SCITEPRESS (Science and Technology Publications, Lda.)

refer to the method as multi-layered scenario analy-

sis (MuLSA). This paper is constructed as follows. In

the next section, we introduce the related work. In

Section 3, we describe the basic concept of MuLSA

and overview MuLSA with its analysis structure and

process. MuLSA is evaluated by applying it to a case.

We describe the case and the results of the application

in Section 4. In the rest of the paper, we discuss the

effectiveness of the method and conclude this paper.

2 RELATED WORK

2.1 Emotion Analysis

Plutchik (Plutchik, 1980) introduces three-

dimensional circumplex model. In the model,

every emotion is composed of the combination of

eight basic ones: vigilance, rage, loathing, grief,

amazement, terror, admiration, and ecstasy. We

selected and categorized these emotions into positive

and negative emotions. For example, ecstasy and

admiration can be categorized into positive emotions.

Thus, vigilance, rage, loathing, grief, amazement, as

well as terror are categorized into negative emotions.

If we map these emotions according to customer

satisfaction, the positive/negative emotions are

reﬂections of/and refer to the high/low quality of the

service for the customer. For example, if the ADAS

can provide safe driving, the driver is satisﬁed with

the service of the ADAS, and will accept the system.

“Acceptance” is a kind of admiration according to the

Plutchik’s three-dimensional circumplex model. On

the other hand, if the driver is fearful of the system,

we can regard the quality of the service as low, and

needs to be improved.

There are several means to measure emo-

tions (Lorr, 1989). We developed a tool to observe the

driversemotions by means of simulation, from which,

we can get quantitative measurements by means of

keys that are assigned to the positive and negative

emotions. Some researchers use questionnaires to as-

sess emotions (Wallbott and Scherer, 1989). We on

the other hand, do not use questionnaire, but employ

a tool and interviews.

2.2 Scenario Analysis

J. M. Carroll (Carroll, 1999) describes the strengths

scenarios. Scenarios can explicitly envision and doc-

ument typical and signiﬁcant user activities. It also

provides us reasoning with regard to the situation of

use before we develop the system. Use case (Jacobson

et al., 1992) and user story (Benyon and Macaulay,

2004) are categorized into scenario analysis methods.

Persona analysis is sometimes used to analyze a spe-

ciﬁc user’s activities (Aoyama, 2005). Even though

a scenario reports a user’s activities well, we need to

analyze more than these reported activities. The prob-

lem with a simple scenario, such as use case, is that

it is constructed in a single-layered structure. In or-

der to evaluate the quality of ADAS, we need a more

complex analysis space.

In the service analysis, there are methods to ana-

lyzing the quality of services by the customer satis-

faction. The concept of those methods is that, “the

customer is satisﬁed with good service.” For exam-

ple, if the customer is unsatisﬁed with the service, it

should be improved.

A customer journey map (CJM) (Stickdorn and

Schneider, 2012) is used to evaluate the customers’

emotion while mapping them to their services. It is

a kind of scenario analysis method. A scenario de-

scribed in the CJM has a time line and a concrete

story. The service receiver accesses the service at

a touch-point. The emotions are presented accord-

ing to the time line of a scenario in the CJM. The

negative emotion of a customer implies some prob-

lems within the services provided. Though there is

no standard notation for CJMs, most CJMs have a

two-layered structure. In the ﬁrst layer, services are

shown as a user story with touch-points. The story

can be regarded as a customer’s journey or experi-

ences in the forest of the services. In the second layer,

the emotions or impressions of the customer are de-

scribed according to the customer’s journey. Richard-

son (Richardson, 2010) shows an example of a jour-

ney into shopping. The scenario commences from a

customer’s awareness to out-of-box-experience. Dur-

ing the customer’s activities, a CJM is used to evaluate

motivations, questions and barriers.

Risdon (Risdon, 2011) proposes a CJM to the

analysis of the service of the Rail Europe. We can

see a lot of examples of CJM on the Internet. CJM

helps service marketing or business marketing im-

prove their services or products. Our purpose is to

analyze the quality and/or problems of the services

provided by the DAS and prioritize requirements of

the future ADAS. Its analysis space contains user’s

activities, user’s emotions, environment of the usage,

as well as services. Thus, our method has a multi-

layered structure. We refer to this method as MuLSA,

which is an abbreviation of “Multi-Layered Scenario

Analysis.”

A service blueprint (SB) is also a two-layered sce-

nario analysis method (Shostack, 1984). This service

is provided to a customer via a front stage of the ser-

vice that is constructed in a back stage. In the frond

ICSOFT2013-8thInternationalJointConferenceonSoftwareTechnologies

stage, direct communication between a customer and

services is shown. In the back stage, there is indi-

rect interaction between the customer and the under-

ground mechanism that supports the services. There

are several relationships between these two layers.

They represent how the services are implemented and

serve to the customer. However, it does not analyze

the emotions of a customer, but simply designs the

services. We integrate a CJM and an SB in order

to analyze priorities of requirements to improve the

quality of services through the application of MuLSA.

Blueprint+(Polaine et al., 2009) is a method that

integrates a CJM and an SB. It is also a two-layered

scenario analysis method. The ﬁrst layer is a cus-

tomer’s layer with three diagrams, i.e. fail line, emo-

tion, and cost. The second layer is a system layer

with a set of actors and a touch-point for each ser-

vice. We can interpret the system layer as an actor

that implements the service through the touch-point.

The strength of MuLSA is that it has a context layer to

represent the environment of the usage, which affects

the services. Blueprint+ does not have such a layer.

The context layer of MuLSA reﬂects the environmen-

tal factors of the services. It helps analyst understand

the context of the service. The context of the service

of the DAS is important, since the service of ADAS

has to be modiﬁed according to the environment of

the car in service.

3 MuLSA: MULTI-LAYERED

SCENARIO ANALYSIS

In this section, we describe the basic concept of

MuLSA and give an overview of MuLSA as a method

of analyzing customer satisfaction and the quality of

the services of the current system.

3.1 Basic Concept

It becomes possible to drive a car automatically. For

example, a driverless Audi TTS climbed up the top of

Pikes Peak in 2010 (Kuchinskas, 2010). Google also

developed a google driverless car (Markoff, 2010).

Our focus is not an automatic driving system. We

consider that although a car can be driven safely by

a computer, the autonomous system will not satisfy

its human driver and passengers.

A car is on a road under various environments and

other cars may not behave expected way. We have

to analyze the environmental and mechanical circum-

stances under which the services perform. Events

and/or objects that are monitored by the ADAS are

referred to as the environmental factors. Some of the

environmental factors are called “hazards.” When we

elicit and analyze requirements of ADAS, we have

to consider the possible hazard and keep the car and

driver out of danger.

Drivers, however, are the experts who detect haz-

ards when they drive their car. When they detect a

hazard, they become strained. If theyfeel they are free

from the hazard, they must be relaxed. When they re-

alize that they are not being cared for by the ADAS,

they become scared and/or irritated. We can refer to

these emotions, i.e. strained, surprised, scared, irri-

tated, etc., as negative emotions. Further, negative

emotions may lead a driver to the dissatisfaction of

the system. Thus, we can detect hazards by monitor-

ing the emotions of drivers and deﬁne requirements

for the future ADAS. We expect to elicit highly pri-

oritized requirements by analyzing real driver’s, and

their emotions with regard to, and caused by the ser-

vice reception from the ADAS.

On the other hand, emotions, such as showing en-

joyment, being relaxed, etc., are referred to as posi-

tive emotions. The mission of the ADAS is not only

to keep a driver and fellow passengers safe, but also to

give positivefeelings to the driver. In order to increase

the satisfaction of a driver, we analyze the services

that dissatisfy drivers, and clarify the reasons drivers

feel negative emotions with regard to the services.

Companies that provide DAS sometimes research

driver satisfaction through questionnaires. Some

companies also survey driver satisfaction for each

country the systems operate in. However, the ques-

tionnaire survey is not adequate for detecting prob-

lematic services. In order to detect problematic ser-

vices, we need to analyze the process of the service

provision and the usage of the service. MuLSA has

been developed to analyze the process and usage of

the service of DAS.

3.2 Overview of MuLSA

MuLSA integrates a CJM and an SB through the ad-

dition of extending with environmental factors as the

context of the services. The analysis structure of

MuLSA is shown in Figure 1. It has three layers.

• Customer’s layer

In this layer, two kinds of information are pre-

sented. One is the customer’s journey, and the

other is the emotional changes of the customer. In

Figure 1, this layer is shown at the top of the ﬁg-

ure. The time line of the scenario is passed from

left to right.

• Context layer

The context is shown in this layer. In the context

MuLSA:Multi-LayeredScenarioAnalysisforanAdvancedDriverAssistanceSystem

Customer’s layer

experiences

emotions

Context layer

E.F.a

Service Mechanism Layer

service_x

(a front stage)

time

positive

negative

E.F.b

E.F.c

service_y

(a back stage)

service_z

(a back stage)

touch point

situation/action

impression/insight

Figure 1: The structure of MuLSA.

of the services, environmental factors that affect

the service contents are identiﬁed.

• Service Mechanism layer

This layer is constructed with two sublayers that

are a front stage and a back stage. The front

stage is a communication facade between cus-

tomers and the system. The back stage represents

major components of the system. This layer is

shown at the bottom part of Figure 1.

3.3 Measurement of Emotions

Psychologists have proposed methods to measure

emotions (Russell, 1989). For example, in the case

of self-report questionnaires, it is the test subject who

reports intuitively and/or subjectively their emotions

expressed through the use of various words, e.g. hap-

piness, surprise, fear, anger, disgust, or sadness. An-

other way of measuring emotions with regard to vo-

cabulary usage is by the application of a response

scale, on which a test subject is observed so that their

facial expressions can be recorded. Our purpose in

measuring emotions is to clarify any problematic be-

haviors with testing systems according to the subjec-

tive emotions of test subjects. Thus, we consider that,

intuitive reporting is not only important, but if possi-

ble, we also expect a test subject to report their emo-

tions consciously, because followingthe simulation or

test, we may ask them what they feel and what they

require of the future ADAS.

A tool to record their emotions is simple. A test

subject only inputs keys according to their emotions.

The positive and negative emotions are ranked into

four levels. The highest emotions are assigned to

a key “a.” Further rankings on our declining scale

are, higher, high and rather positive, which are as-

signed to the keys “s”,“d”,“f”, respectively. Similarly,

worst, worse, bad, rather negativeare assigned to keys

“l”,“k”,“j”,“h”, respectively. Hence, the test subject

can set their hands on a keyboard and type the char-

acters according to their emotions. The tool is simple

enough and needs little training. Moreover, the tool

records the timing of the key ins and the key itself, and

interprets input keys to the emotional scale described

above, from -4 to +4. As a result, we can establish

an emotional evaluation of the test subject as shown

in Figure 2. In our next version of the tool, a joystick

may be applied as an input device.

(+)Satisfaction

(-)Dissatisfaction

time

(sec)

Emotional

scale

emotions

Figure 2: The emotions monitored via key ins.

3.4 The Requirements Analysis Process

With this three-layered scenario, we can gather the

points of the emotional changes and analyze the state

of the DAS and the environment. In order to develop

the ADAS, we have to analyze problems within the

DAS, which has already seen application in the mar-

ket. The ADAS has to be able to analyze the circum-

stances and environment of the car through the use of

numerous sensors. The requirements analysis process

of MuLSA is shown below.

1. Identify subsystems and set them in the back

stage.

These subsystems represent the limitations of the

current system. The sensors and controllers of the

DAS can be detected. Deﬁne the functionalities

and efﬁciency of those components according to

the real components in the current system. In or-

der to elicit requirements for the future system,

we need to visualize the limitations of the current

system.

2. Identify components in the front stage.

A display is the most typical component in the

front stage. Further, alarms, beeps, or announce-

ments can be components as well.

3. Identify hazards as the environmental factors.

In order to assess the quality of the services of the

ICSOFT2013-8thInternationalJointConferenceonSoftwareTechnologies

current system, listing hazards with regard to en-

vironmental factors as much as possible is done.

Hazards will be the context of the scenario deﬁned

in the next step. Weather, road condition, periph-

eral vehicles, pedestrians and/or animals can be

the hazards.

4. Construct a scenario as a customer journey.

The scenario is deﬁned to analyze how much a

given hazard affects a driver’s emotions. If we can

know how much the environmental factors affect

the drivers’ satisfaction, we can prioritize sensors

that can detect and monitor the environmental fac-

tors. The best length of the scenario is still under

consideration. The scenario of our ﬁrst case is less

than 30 seconds.

5. Develop simulation.

The scenario is transformed into animation that

is developed with PreScan (Advanced Simulation

Technologies Ltd., 0132). PreScan is a develop-

ment environment for ADAS or intelligent vehi-

cle systems. Two kinds of animation must be

constructed as scenes seen through the windshield

and a rearview mirror. Figure 3 represents the im-

age of a tool with the movie being developed by

PreScan.

6. Simulate the scenario with a test subject and get

emotional data in the scenario. After the simu-

lation, we map the emotional data in accordance

with the simulation.

7. Analyze the emotional data and elicit require-

ments for the ADAS.

ACC is one of the current DASs. ACC safely controls

the speed of the car, while establishing a safe distance

from the precedent car. Sensors send signals to the

ACC, which in turn sends other signals to the speed

control unit and display. The future ADAS will have

more sensors and be able to establish the driving con-

text with regard to the environmental factors.

ACC

The image in the

rearview mirror

The image seen

from the

windowshield

ACC

status

start

stop

Emotion monitor

Figure 3: The tool image of MuLSA.

4 CASE STUDY

4.1 Overview

This section describes a case study by which we de-

tect problems in the current ACC as an example of the

DAS, which we then analyze in order to evaluate the

effectiveness of MuLSA. There are various services

on ACCs. The ACC used in this case study provides

the following services.

• The ACC starts when the driver turns on the ACC.

• The ACC is terminated when the driver turns off

the ACC.

• The driver can set the speed for the ACC.

• The driver can increase or decrease the speed of

the cruise within the permissible range.

• If there are no precedent vehicles or, there is

enough distance from the precedent vehicle, the

ACC maintains the set speed of the cruising car.

• If there is a precedent vehicle, the ACC keeps the

adequate distance from the precedent vehicle by

adjusting the cruising speed. The precedent vehi-

cle is detected by a radar censor on-board.

• If the windshield wipers are used in strong mode,

the ACC is automatically terminated, so that the

radar or laser cannot detect the precedent vehicle.

• If the precedent car goes out of its lane, the ACC

gradually turns the speed back to the set speed.

• If the speed of the cruising car becomes slower

than a certain speed, the ACC is automatically ter-

minated.

The display in which the state of the ACC is

shown is identiﬁed in the front stage, with the sen-

sors, speed controller and ACC being deﬁned in the

“back stage”. In order to evaluate MuLSA, we made a

simulation with PreScan(Advanced Simulation Tech-

nologies Ltd., 0132). We took various environmen-

tal factors into account within the simulation. They

were weather condition, a precedent vehicle, as well

as another car that cuts in front of the cruising car.

A test subject who is a driver accesses the simulation

via a keyboard and display interface through a per-

sonal computer. The insights of the test subject were

monitored by their utterances during the simulation.

In our future tool, the emotions of the test subject will

be automatically recorded from the keyboard.

4.2 The Scenario

The test scenario is as follows.

MuLSA:Multi-LayeredScenarioAnalysisforanAdvancedDriverAssistanceSystem

1. The driver increases the speed up to the desired

speed and turns on the ACC.

2. The ACC comes into service state and starts to

provide its services with the car cruising at the de-

sired speed.

3. The driver releases the accelerator pedal.

4. A vehicle cuts in front of the car. Then, the sensor

detects the vehicle and alarms the distance to the

ACC.

5. The ACC decreases the speed of the car in order

to keep an adequate distance from the precedent

vehicle.

6. The driver feels the sudden gravity of reducing

speed.

Since the test subject only watched the simulation

movie, the change in gravity was communicated

to the test subject from the staff. The event of the

“reduced speed” was caused by the precedent ve-

hicle, which is one of the environmental factors,

and which is dispatched via sensor and the ACC.

7. It starts to snow heavily.

8. The driver turns on the windshield wipers to the

strong mode to keep visibility.

9. The ACC catches the event.

10. The ACC terminates its services to avoid sensor

errors and notiﬁes the driver of the termination via

the display.

11. The speed of the car is reduced: the result of

which sees the following vehicle increasing its ap-

proach.

The simulation is made on the assumption that the

driver can notice the termination of the ACC from

the display. If the test subject does not notice the

situation, the staff informs the tested of the situa-

tion. The simulation is a movie, so the test subject

is not actually operating the car.

12. The driver notices the termination of the ACC.

13. The driver puts their foot on the accelerator pedal

and restarts the manual driving.

The results of MuLSA is shown in Figure 4. The emo-

tions of the test subject and their journey are shown

in the customer’s layer. Their insights, recorded

from their utterance, are shown in balloons. There

were two environmental factors: other vehicles and

weather. These factors are shown in the context layer.

In the next section, we analyze the results.

4.3 Requirements Analysis for the

ADAS

MuLSA consists of three layers. We can see the test

subject’s touch-points through the ACC, as well as

when and how strong the test subject (driver) had pos-

itive or negative emotions toward the system (ACC)

within MuLSA’s layered structure. We consider that,

the negative/positive emotions of the test subject may

imply problematic and/or ideal behaviors of the cur-

rent ACC. In order to analyze in detail the emotion

in and of each touch-point, we refer to the recorded

utterances of the test subject. They are shown as bal-

loons in Figure 4. For example, the ﬁrst balloon (A)

represents the feeling when the test subject displayed

the positive emotion. This is the effect of the service

(2).

The purpose of our research is to analyze require-

ments of the future ADAS. Hence, we focus on the

negative emotions in Figure 4 and refer to the causes

of the test subject’s negative emotions. The causes

must be shown in the context of the scenario. If the

context of the car is correctly detected by the ACC,

the problem is in the ACC software. If the context of

the car is not correctly detected by the sensors of the

ACC, we have to consider the addition of new sen-

sors to the future ADAS. The priority of each sensor

can be set in accordance to the level of the negative

emotions.

The second balloon (B) represents the feeling felt

when the test subject detected the sudden gravity

change. Though they must have been surprised at the

change in gravity, they thanked the ACC for avoiding

danger, in this case, a car crash. We regard such a sur-

prise may decrease the customer satisfaction. This is

a problem with the software of the ACC itself. The

future ADAS may require smoother speed control, so

that the driver does not feel fear or surprise.

The problems that we have to solve lay within the

balloon (C). The test subject did not understand the

speed decline. This means that the ACC must com-

municate its state securely to the driver. However, the

ACC had been terminated due to the snow, thus, the

test subject noticed the state change securely. If the

test subject knows that the ACC may be stopped in the

inevitability of low visibility, they could prepare for

the termination and would not have the negative emo-

tion. We need to redesign the behavior of the front

stage.

So far, we have been able to elicit two new re-

quirements for the future ADAS. One is the smooth

speed control, in which case, the ADAS does not al-

low a driver to feel a sudden gravitational change. In

order to satisfy the requirement, we need to analyze

the risks with regard to the safety of driving as relates

to gravitational changes. The other is that of the com-

munication between the driver and the ADAS. Some

kinds of announcements may annoy drivers. How-

ICSOFT2013-8thInternationalJointConferenceonSoftwareTechnologies

Start the

service

Driver’

experience

Other

vehicles

Weather

in service

It snows.

Detect

the

speed

declines

Felt the

sudden

gravity

Alarm

the

distance

Decrease

the speed

Alarm of the

possibility of

sensor errors

ACC

Display

Sensor

Speed

Controller

Start

Visibility

becomes

low.

I got a foot

off an

accelerator

and was

relaxed.

Oops!

I was almost

crashed with

the precedent

car.

Whoa! There was

enough distance

between my car and

the precedent car, but

the car following me

has come too closely.

What happened?

Switch on

the

windshield

wipers.

Gravity

falls, and

the distance

between the

cars has

opened.

Release

the

accelerat

or pedal

Opened

the

distance

(+)Satisfaction

(-)Dis-

satisfaction

Notice that ACC

has been

terminated.

Started

the

manual

drive.

Emotions

(A)

(B)

(C)

(1)

(2)

(3)

(4)

(5)

(6)

(7)

(8)

(9)

(10)

(11) (12) (13)

Customer’s

layer

Context

layer

Service

Mechanism

layer

Front Stage

Back Stage

Service State

Monitor

the

distance

out of service

Display

“ACC

terminated”

Increase

the

speed

Thank

you,

ACC!

Oh, how

disappointing

0:22:20

0:24:07 0:26:03

0:41:11

0:43:13

0:44:19 0:56:04

0:15:07

time

(sec)

A vehicle

cuts in front

of my car

eather

Othe

Emotions

Figure 4: The analysis with MuLSA.

ever, important messages, e.g. start, termination, etc.,

have to be securely communicated to the driver. For

example, if the driver does not notice the termination

of the ADAS, a following car may run into their car.

The balloon (C) tells us that the driver realized the

possibility to be of such a situation. The priority of

this improvement of the notiﬁer within the display is

high.

This case study shows us the effectiveness of

MuLSA, because we were able to elicit new require-

ments and their priorities.

5 DISCUSSION

AND CONCLUSIONS

MuLSA is a satisfaction basis method. In this paper,

we elicited new requirements of the future ADAS by

applying MuLSA to the current ACC.

We are now developing other scenarios with more

vehicles and various road conditions. In these future

scenarios, we will put the ACC into the various haz-

ardous situations, to be solved by the ADAS.

We were able to evaluate the effectiveness of

MuLSA by the case study. However, MuLSA is not a

method only for the ACC or ADAS.

As Kimbell (Kimbell, 2011) described, one of the

strengths of the user stories is that they proposes ideas

for new services components and also entirely new

services. MuLSA is also applicable to most soft-

ware, when an analyst needs to elicit new require-

ments based on the current software. The strength of

MuLSA is that it analyzes requirements through the

utilization of its multiple-layered structure in which

there are users, context of the usage, as well as the

mechanism of services. Researchers have proposed

MuLSA:Multi-LayeredScenarioAnalysisforanAdvancedDriverAssistanceSystem

a lot of methods for requirements elicitation, such

as goal oriented analysis methods (Dardenne et al.,

1993; Yu, 1997) or use case analysis (Jacobson et al.,

1992) focuses on initial requirements elicitation for

new software. In contrast, MuLSA focuses on soft-

ware that is developed as an innovation on the current

software.

The weakness of MuLSA is that its analysis pro-

cess is not so systematic. It depends heavily on

the emotions and/or insights of users, rather than the

goals or purposes. MuLSA is a kind of scenario anal-

ysis method. The scenario provides a real story within

time. As Carroll mentioned (Carroll, 1999), scenario

is understandable for every user and gives a real expe-

rience to them. New requirements for innovations on

current software are hard to elicit through interviews.

We believe that most important requirements must be

elicited from the users’ real voice or emotions as a

result of their experiences, rather than requirements

analysis work based on a table.

An analyst with MuLSA does not expect the users

to proposeproblems or new requirements, rather, their

emotions and insights in their use of the current sys-

tem is key. The effectiveness of MuLSA is to ana-

lyze the causes of the users’ emotions. As a result,

we can prioritize new requirements for the software

of the next generation. This means that the scenario

has to contain situations in which the user realizes the

problems of the current system. For example, it can

be used for the claim analysis (Carroll, 2000) which

needs to analyze various users and usages.

In this paper, we proposed a method named

MuLSA to elicit requirements and prioritize them ac-

cording to a scenario analysis. The method is being

developed for the improvement of future software as

the next generation of current software or systems.

The scenario has multiple-layers, with the customer’s

layer, context layer, as well as the service mechanism

layer. The customer’s layer can be used in claim anal-

ysis for various users. We are able to deﬁne negative

actors in the context layer. It is efﬁcient to analyze

misuse cases and/or analyze requirements under var-

ious situations (Alexander, 2003). The multi-layered

structure with time, makes it possible to analyze mis-

use cases more effectively than through a use case di-

agram.

In our case, we decompose the mechanism layer

into several sublayers. If we apply MuLSA to a gen-

eral software analysis, the mechanism layer may need

two sublayers, i.e. a front stage and a back stage.

ACKNOWLEDGEMENTS

The authors thank Ms. Mineko Naoe for developing

the tool to monitor emotions.

REFERENCES

Advanced Simulation Technologies Ltd. (2013/2). PreScan,

http://www.advancedsimtech.com/software/prescan/.

Alexander, I. (2003). Misuse cases: Use cases with hostile

intent. IEEE Software, 20(1):58–66.

Aoyama, M. (2005). Persona-and-scenario based require-

ments engineering for software embedded in digi-

tal consumer products. In IEEE International Con-

ference on Requirements Engineering, pages 85–94.

IEEE Computer Society.

Benyon, D. and Macaulay, C. (2004). A Scenario-based De-

sign Method for Human-Centered Interaction Design,

pages 211–235. John Wiley & Suns.

Carroll, J. M. (1999). Five reasons for scenario-based de-

sign. In The 32nd Hawaii International Conference

on System Sciences.

Carroll, J. M. (2000). Making Use: Scenario-Based Design

of Human-Computer Interactions. MIT Press.

Dardenne, A., van Lamsweerde, A., and Fickas, S. (1993).

Goal-directed requirements acquisition. Science of

Computer Programming, 20:3–50.

ISO/IEC 25000:2005 (2005). Software engineering – Soft-

ware product Quality Requirements and Evaluation

(SQuaRE) – Guide to SQuaRE.

Jacobson, I., Christerson, M., Jonsson, P., and Overgaard,

G. (1992). Object-Oriented Software Engineering.

Addison-Wesley.

Kimbell, L. (2011). From novelty to routine: Services in

science and technology-based enterprises. In Design

for Services, pages 105–111. Gower Publishing.

Kuchinskas, S. (2010). Automatic auto: A car that

drives itself. http://www.scientiﬁcamerican.com/

article.cfm?id=self-driving-audi-automobile.

Lorr, M. (1989). Models and Methods for Measurements

of Mood, volume 4 of The Measurement of Emotions,

chapter 2, pages 37–53. Academic Press.

Markoff, J. (2010). Google cars drive themselves, in trafﬁc.

http://www.nytimes.com/2010/10/10/science/

10google.html?pagewanted=all&

r=0.

Plutchik, R. (1980). Emotion - A Psycho-evolutionary Syn-

thesis. Harpercollins College Div.

Polaine, A., Aebersold, R., Bossart, R., and Mettler,

A. (2009). Blueprint+, http://servicedesigntools.org/

content/148.

Richardson, A. (Nov.15.2010). Using customer journey

maps to improve customer experience. Harvard Busi-

ness Review, Blog Network.

Risdon, C. (2011). The anatomy of an experi-

ence map, http://www.adaptivepath.com/ideas/

the-anatomy-of-an-experience-map.

Russell, J. A. (1989). Measures of emotion. In Plutchik, R.,

editor, Emotion -Theory, Research, and Experience,

ICSOFT2013-8thInternationalJointConferenceonSoftwareTechnologies

volume 4 of The Measurement of Emotions, chapter 4,

pages 83–111. Academic Press.

Shostack, G. L. (1984). Designing services that deliver.

Harvard Business Review, 62(1):133–139.

Stickdorn, M. and Schneider, J. (2012). This is Service De-

sign Thinking: Basics, Tools, Cases. Wiley.

Wallbott, H. G. and Scherer, K. R. (1989). Assessing emo-

tion by questionnaire, volume 4 of The Measurement

of Emotions, chapter 3, pages 55–82. Academic Press.

Yu, E. S. K. (1997). Towards modelling and reasoning sup-

port for early-phase requirements engineering. In The

3rd International Symposium on Requirements Engi-

neering (RE’97), pages 226–235. IEEE.

MuLSA:Multi-LayeredScenarioAnalysisforanAdvancedDriverAssistanceSystem