Development of Humanoid Robot Design Process

Focused on the concurrent engineering based humanoid robot design

Kwang-Myung Oh, Ji-Hoon Kim and Myung-Suk Kim

Korea Advanced Institute of Science and Technology (KAIST), Department of Industrial Design
[email protected] [email protected] [email protected]

Abstract: This paper presents a design research conducted as a part of a larger project on
intelligent humanoid robots. The primary goal of this study was to develop an initial understanding
of how design on humanoid robots should be carried out.
Humanoid robot design is different from conventional industrial design. A humanoid robot needs to
have personality and character because it interacts emotionally with humans. Also, the design
process should differ from that of machines such as automobiles and airplanes because it moves
autonomously by recognizing its environment. As the development of human-computer interaction
and interface design owe to the growth of computer industry, humanoid robot design should be
studied widely and be accepted as a new design research area.
There is, however, few research approaches on humanoid robot design. In order to build a
fundamental study, we have developed new design processes of humanoid robot. They are role-
character design process and concurrent engineering process. The former is used to determine the
basic specifications of humanoid character, and the latter is applied to build the physical humanoid
structure. In order to verify the processes, we have carried out several types of humanoid robot
design. Through this research we found that the new design process could effectively solve the
traditional conflict between the design part and the engineering part. This paper presents our design
process and proposes initial guidelines for the design of humanoid robots.

Key words: Humanoid robot design, role-character design, concurrent engineering.

1. Introduction
1.1 Background of Research
Humanoid robot, intelligent robot with anthropomorphic form, is regarded as one of the most important
growth engine technology in national economy and its application is also expected to be promising industry in
the new generation after the car industry dominant generation.
Currently developed humanoid robots such entertainment and personal service robots are required to
satisfy a lot of complex conditions such as height, weight, intimacy, safety and flexibility and with these reasons
the approach of product design start to be applied in humanoid robot development.
Humanoid robot design includes action and behavior structure design and human-robot interaction design

1
as well as appearance design. Especially for personal service robot which assists and supports human life in
domestic environment it is more important to have emotional exchange with human and personality design for
humanoid robot is one of the essential issues.
Through this integrated process, humanoid robot design can play an important role as an interface between
human and high technology and this research on humanoid robot design process can reveal new possibilities of
solution to solve the problems of current technology-centered robot development.

1.2 Research Outline

In this research, two different processes for humanoid robot design are suggested. One is the design
deployment based on role-character implantation and the other is application of concurrent engineering design
method. After completion of both design processes, through the design visualization process, design
development is finalized.
1) Role-character based humanoid robot design (role-character design process)
For higher level of interaction with human, humanoid robot is required to have specific type of personality
because human tend to treat another human with specific personality. Thus in the design development of social
humanoid robot, personality design and step wised process for this is required. Firstly the classification of
existing robot according to role and environment is explored and analyses on appearance and interaction
behavior are conducted. Secondly scenario method is employed to design specific personality and role of a robot.
Scenario method is applied to predict and visualize robot in simulated situation and for priority measurement,
MSO methodology is used. After the completion of scenario design, specific personality and roles are decided
and the results are transmitted to concurrent engineering design process.
2) Concurrent engineering for humanoid robot design (concurrent engineering design process)
The appearance of a humanoid robot consists of frame structures which function as skeletons and housings
For frame components which support weight of humanoid robot and generate movement, very precise
engineering approach is required but for housing components, industrial design concerned-approach is
indispensable because appearance aspects function as emotional man-machine interface.
To combine two different properties and approaches more efficiently, concurrent engineering method is
required. In concurrent engineering process for humanoid robot design, scenario and questionnaire which is
made up of visual images play an important role as a communication tool between design team and engineering
team.
3) Visualization of design
After the iterative process of the previous steps, appearance design is refined and finished with image
mapping for the visualization stage and through the final turn of questionnaire and scenario, structural defect and
confliction is modified and corrected in accordance with both sides.
In the line with this, for humanoid robot design development, concurrent engineering design approach is
indispensable and design team also need to participate in the project from the bottom stage of development
process.

2
2. Robot Design Process
1
Bartneck, robot design researcher, proposed properties of social robots in his literature: form, modality,
social norms, autonomy, and interactivity. Based on the Bartneck’s research, we classified robot design elements
into three groups for the humanoid robot design: character design, appearance design and interaction design.
Character design is the designing of properties and status felt from the type of feature as an example of
‘Pino(2000, Japan)’ motivated by Pinocchio. In this stage, designer defines the role and characteristic of robot
and then, builds a design scenario to represent detailed personality. The scenario methodology make robot
technology can operate effectively and friendly to human.
Appearance design is the designing of the exterior view style of a robot. In this step, designers should
consider the engineering specification for practical performance and the use and environment defined by role and
character of robot. This design stage includes structure as well as mechanism, shape, material and color.
Besides, interaction design is a process of finding the robot’s appropriate behavior model and designing
visual, auditory and haptic interaction. To realize easy, speedy and comfortable interaction, simulation technique
like a software-robot (Sobot) is needed

Humanoid robot design process

Role-Character Design Concurrent Engineering Visualization of Design
1. Robot classification as role and 1. Determination of robot’s basic 1. Image mapping and adjectives
environment. specification by 1st questionnaire. extraction
2. Setting up the direction of 2. Extracting refined design 2. Concept and Idea sketch
character design. specification by 2nd questionnaire. 3. Modeling and dynamic structure
3. MSO scenario technique. 3. Adjustment between the designing test
4. Establishment of detail roles and and the planning of frame. 4. Final design solution
character.

<Fig.1 Humanoid robot design process>

These design elements are correlative and fulfilled in consecutive order. In this research we gave priority
to the basic robot appearance, thus character design and appearance design were conducted with the excluding of
interaction design. The processes of character design and appearance design are treated in role-character design
stage and concurrent engineering stage shown in <Fig.1>.

2.1 Role-Character Design Process

Each person has a specific type of personality. If a passive person speaks actively and behaves
energetically, many people would feel embarrassed and uncomfortable toward him. People tend to treat others
with their specific personalities, so should humanoid robots which have human-like appearances should have
static characteristics.

2.1.1 Classification of robot as role-character and establishment of design direction.

For establishing humanoid robot’s characteristics, we classified humanoid according to several features:
robot’s role, using space, properties of appearance.

1
Christoph Bartneck, 2004, “A Design-Centred Framework for Social Human-Robot Interaction”, Department of Industrial
Design, Eindhoven University of Technology

3
 1) Classification by living territory and roles:
According to the 2 research of JARA(Japan Robot Association) and IFR(International Federation of
Robotics), robots could be grouped in home, industry and public place by living territories. And they could be
classified in five by the roles: entertainment, information, help, security, chores. <Table 1>

<Table 1. Classification by living territory and roles>

Design Direction

Human Home Work Public

Asimo, P3, Aibo, MIT robot ‘Pong’ Asimo
Entertainment
Papero

Information Cooking Robot, R100 Mimizku, S-doll

Help TMSUK4, Dama, Construction Robot, TMSUK4, Robodoc,

HAL9000, Posy Davinci

GuardRobo, GuardRobo, Fire Robocue, Fire-

Security
Monitoring Robot search Fighting Robot
Food handing Robot, Unimate, Maymay Wall-painting Robot,
Chores
Robot Cleaner Trash-Cleaner
Machine

In the point of robot’s role, tasks related with security and chores are suitable for differentiated form better
than humanoid form. And in the point of robot’s territory, very high level of interaction which would be
impossible to embody with current technology is needed to home robot. Accordingly we focused on
characteristics of ‘Infotainment (information + entertainment) robot for public use’ as the first step. And the
technology getting more developed, the direction of humanoid design could move from public robot to home
robot.
2) Classification by physiognomy:
3
Physiognomy is a study on the character or personality by face-looking. Cognition of behavior model and
personality varies with the shape of a face. For the consistency between face-looking and characteristics of
humanoid robot, we applied physiognomy to humanoid robot design. There are representative eight shapes in
physiognomy: inverse triangular type, oval shaped type, round type, square type, rectangular type, triangular
type, trapezoidal type, elliptical type.
In this research, we took two types, a round face type which stands for sociable and passionate personality
and a square face type which stands for strong-willed and staunch character.

2
Provisional Classification of Service Robots by JARA, 2005
3
Min-ho Kim, 2003, “A Study for Character Expression Method based on Physiognomy”, Proceedings of the Korea Society
of Design Studies Conference, pp 98-99

4
<Table 2. Examples of Classification by physiognomy>

3) Types of character
To apply stationary character pattern like human to humanoid robot, we studied character types on MBTI
(Myers- Briggs Type Indicator) which is representative classifying test of humane characters.

<Table 3. Examples of the Myers-Briggs personality type indicator (MBTI)

Scales Preferences Activity

Extraversion (E)-Introversion (I) Where you like to focus your attention Energy orientation

Sensing (S)- Intuitive (N) How you perceive information Perceiving

Thinking (T)- Feeling (F) How you make decisions Judging

Judging (J)- Perceiving (P) Life style you prefer to adopt Lifestyle

Through the questionnaire to design and engineering team, we fixed ENTP character pattern to our
humanoid robot. ENTP stands for extraversion and intuitive.
4) Classification by etcetera:
The gender of humanoid robot is an important design factor. But there are many disputes regarding how
the gender of robots reflects the social-cultural common idea and how much it is necessary to operate. For few
related researches and dimmed logics, designer’s subjective view intervenes in that matter. In this research, we
selected neutral gender near the male in aspect of strong and stability.

2.1.2 Design scenario according to MSO techniques

Scenario methodology is employed to design specific personality and role of robot. According to 5W-1H
principle (Where, Whom, When, What, Why, How), we divided robot situations to three level of MSO: Must
(robot’s minimum duty of basic role which must be done), Should (additional element to enhance elasticity),
Ought to (applicable element as techniques developed afterward). Through this methodology, we could
determine the priority between the present condition of techniques and requirements and set up hierarchical
structure by robot’s role.

5
2.2 Concurrent engineering design process
Intelligent humanoid robot will be the result of interdisciplinary research which extends over long time. In
this research, contrary to widely used sequential process such as first engineering- second designing approach we
took the concurrent engineering method which could help smooth communication between design team and
engineering team. <Fig. 2>

* Before physical planning of robot, design team

determines roles and character of robot. According to Robot Design
that, design team makes design scenario of robot.

*Role-Character Appearance Design Set-up Design Development¹

Communication through the

questionnaire and scenario image.
Frame & Form & Detail Design
Support Structure Exterior Structure
Bench-marking Function Deployment Adjective Design
Form Analysis Partial Proportion
Image Mapping simulation
Joint Expression
Behavior Analysis

System Func- Risk &

Feature Structure Structure Proto- Motion
Integration tional
Spec. Analysis Analysis Modeling typing interference
²
l i

Experimental Drawings Stable Drawings Production

Schemes ( initial planning ) ( detailed planning ) Released
Time ( Progress )

1. Design Development : As the determined appearance design specification, design team develops design agenda by sketching and modeling
2. System integration : To setting up the technical specification of robot, engineering team integrates hardware part and software part

<Fig.2 Concurrent engineering chart for humanoid robot design>

In this process, we progressed considering the appearance properties in design part and frame properties in
engineering part. Firstly in engineering team, they gave the extra shape space to final level of design by creating
maximum available space in structure planning stage. And for the next in design team, we suggest expected
robot appearance image to engineering team, so we help them to apply design image to their building structure
process.

2.2.1 Analysis of humanoid robot component

For the concurrent engineering process, robot designer need to grasp robot structure planning in
engineering team. And at the same time, it is required for designer to analyze of composition element of robot
which influence on design process. In this research, we searched for the outline of humanoid robot structure and
movement mechanism. After that, we arranged the important component concerned with design process:
proportion, position of battery, visualization of articulation joint. <Table 4>
Components of robot are divided as research parts for the concurrent engineering process and according to
the <Table 4> communication with engineering team are preceded.

6
 <Table 4. Analysis chart of robot component >

Appearance Factor Research part Class Component of robot

Battery Battery position according to Mechanic

Structure position structure
& Engineering (inside or outside, back, breast, leg)
Mechanism Joint Presence of hip and waist ,
structure Line of rotation flow and rotate axis

Life size
Body Bust : legs
Engineering proportion Leg : arm
Ratio of front side view
+
Shape Design Division of upper body axis
Visualizatio
Shoulder joint according to DOF
n of joint Neck, knee, elbow, hand, food

Shape of face
Face image Visualization of feature
Design Impression of whole shape
Cosmetic factor Color, material

2.2.2 Communication through questionnaire and scenario image.

The most important thing in concurrent engineering design process is the procedure to integrate
engineering proposal and design idea. The planning processes of engineering team generally tend to focuses on
the partial solutions and are decided by superficial observation because robot development is composition of so
many sophisticated and independent technologies. For solving this problem and converge engineering proposal
upon design process efficiently, designer have to get opinions of engineering team through the 4visual image by
presenting partial design suggestion in questionnaire. The restriction of engineering team is concerned with the
numerical problem, so designer could take the concrete definition of engineering team by visualizing numerical
value or proportion of humanoid robot in communication.
For the communication we made questionnaire sheet based on the previous analysis chart of robot
component <Table 4>, and variable proportion or visualization investigated by partial sketch were applied to that
shown in <Table 5>.
In the result, we took the concrete specification of appearance design, and additionally we required
acceptable behavior scenario of engineering team which could be developed in the near future to understand
technical restriction and possibility. This scenario was the final procedure about how robot design reflects the
questionnaire result. And we examined the partial engineering suggestion with that. We seized the engineering
scenario as the four research part: heat, arm, leg, integration.
Scenario 1 was that robot walks to hand in flowers and to shake hands. Scenario 2 was that robot shift cart
together when human call robot and scenario 3 was that robot read e-mail by PDA command. Scenario written
by engineering team and previous questionnaire results is the core to decide design element, so we set up
guideline of design visualization of robot on the basis of that.

4
Jose RIVERA-CHANG, “Use of Scenarios as a Collaboration Tool between Industrial Design and Engineering Students”,
Industrial design California State University

7
<Table 5. Questionnaire and the result >

Question Result and direction

1) 2) 3)

어깨~팔꿈치(18cm),
•Head : body = 1 : 6 팔꿈치~손목(20cm)

•Head : body

어깨~팔꿈치
•Bust : leg = 2: 3

18 CM
•Bust : leg

키= 120 ~150 cm
4) 5) 6)

•Upper leg : lower leg

상반신:하반신
•Upper leg : lower leg

2:3
•Shoulder ~arm = 19cm

허벅지:장단지
Proportion 7) 8) 9)
•Arm~ wrist = 20cm

1 :1
•Smooth, rounded
•Smooth, rounded
Surface •Straight, solid
•Modified round
considering inner spec
a b

•Hidden joint
•Visualized •Soft elements
Joint type •Hidden •Solving shape a b

to working in design developing

•Finger joint
•Finger joint •Simplification level
Hand
•Simplification level •Foot joint a b

•Foot joint •Foot : top A + bottom B

Foot =>Easy to walk a b

•Closed to bust
(Hip joint) •Hip joint
•Closed to leg •Able to rotate
Hip (Not hip) for flexibility
a

•Back •Mainly focus on back

•Chest •Other
Battery •Other (Hop, leg, - considering the center of gravity)

2관절
회전 1관절
1관절

•Necessary points •Arm joint(6 degree of freedom 회전 1관절

1관절

Other by technology constrains •Enough space in bust

suggestion

2.3. Visualization of Humanoid Robot Design

Through the role-character design process and concurrent engineering design process, we took the
visualization of design. The process of visualization is composed of image-mapping, extracting concept, sketch
and 3D modeling.

2.3.1. Image mapping

Based on research, robot image is divided into joint, color, whole image, and face. In detail, joint and color
is divided into machine and toy style, while the whole image and face is divided into animal and human type.
Existing robots are grouped into four regions by the above factors. <Fig.3>

8
2.3.2 Concept making
We defined humanoid robot design concepts into three types:
1. Concept 1 pursues the image that is mechanical in joint shape and color, and humane in the whole image
and face. Robots in this concept have strong mechanical impression, and they are built with straight
lines.
2. Concept 2 has a toy image in the joint shape, color and face. Robots in this concept have curvilinear
shapes, and could have child-like looks by the colors.
3. Concept 3 has an animal and toy shape. Robots in this concept could be developed to mascot-like image.

Joint, Color
Machine

Concept
1

Animal Human
Whole Image,
Face

Concept Concept
3 2

Toy

<Fig. 3 Image mapping>

2.3.3 Abstraction of adjective words

Abstracting adjective words is a useful method for designers to visualize their idea in the idea sketching
process. This step has various advantages for understanding, expanding concept and communicating with other
fields.

<Fig.4 Adjectives of concept 1 and 2>

We focused on expressing of each joints, one of the most important design point, especially at shoulders
and knee parts. Good joint shapes shows the rotating lines of flow, so people could predict the movement.
The face image is a major factor because it decides whether people feel familiar with the robot or not.
Concept 1 is neutral gender near to the male, strong, and simple mood. Concept 2 is a round, cute and child-like

9
looks. Concept 3 is a humanoid, but has characteristics of an animal such as a cat or dog.
Silver and white colors which are popular and acceptable in robots were applied on all concepts. Pastel
tone or vivid colors were used in concept 2 and 3 as accent points.

2.3.4 Idea Sketch

Ideation and sketching was performed on each concept after defining the concepts and directions in detail.
Firstly, we made rough sketches based on the requirement of engineering team. And next, we applied our
sketches into previous rough sketches because of the unification between techniques and design. Besides,
concept 3 could be developed to have marked individuality and emotionality. So we excluded concept 3 in this
research but we planned to take concept 3 in Software-robot development after this research.

2.3.5 3D Modeling
On the basis of idea sketches, we moved to modeling process. Prior to the modeling the final sketches
defining the size and proportion –height 1400 mm, width 550mm – were selected as <Fig.5>. The final sketches
were slightly modified based on the results of questionnaire by the engineering team. After this process, the size
and proportion were modified continuously through the engineering team’s test.

<Fig.5 Developed sketches>

3D modeling of type A and B were completed based on the final sketches. Alias 10.0, 3D software tool,
was used for its modeling and it was modeled by proportion rather than by exact dimensions.
Type A was rendered based on the concept 1 and designed in a straight and mechanical form. And the
facial expression, the most difficult part in humanoid robot design, was designed as a metaphorically expressed
goggle form. Silver and dark grey colors were used to give the simple and strong impression. <Fig.6>
Type B was rendered based on the concept 2 and designed in a round and sleek form. White and blue
colors were applied as accent.

10
 <Fig.6 Rendering of type A and type B>

3. Robot Design Process Evaluation.

We tried to measure the effectiveness of our robot design process. The effectiveness of the process
implicates how well the process can solve the chronic problems between design part and engineering part, and it
could be discussed in two points of view5: design quality, design cost.

<Table 6. Comparison chart of robot design process >

Robot Quantity of Communication Communication Starting Time of Specifics of
with Engineering part Tool Design Participation Process
KHR-3 in 2 times a week. Each Some 2D sketches. After Completion of Limited
KAIST 2hours. Frame Construction appearance styling
works only.
NBh-1 in 2 times a week. Each Image Map. After Completion of Limited
KIST 5hours. Adjective words. Frame Construction appearance styling
works only.
Robot in this 1 times a week. Each Visual Image From the beginning Application of
research 2hours. Questionnaire. of Whole Process DFM and DFA

Design quality represents the amount of designer’s intentions reflected to the output. The more intentions
are reflected, the more the process is effective. Design cost includes the time spent and expense for designing
and revising the design till the final output. If the design cost is in low level, the process can be regarded as high
efficient one.
To evaluate our process, we conducted a focus group interview with two representative robot designers in
Korea (KHR designer and NBh designer). And we compared each design process. First of all, we grasped each
robot design process and caught out different stages and distinctions in each process, shown in <Table 6>. And
for estimating design quality and design cost, we deployed evaluation items based on robot design element.
Through the analysis of each items, each process was evaluated and visualized in diagram shown in <Table 7>.

5
Kwan-myung Kim, 1994, “A Study on the Application Concurrent Engineering Concept in Industrial Design Practice”,
KAIST

11
<Table 7. Evaluation chart of robot design process >

Criterion low, little high, much,

early late
Evaluation Items
Design Quality Initial Robot Design Concept
(Design Proportion and Balance
Contribution to Shape and Position of each part
Robot Color and Material
Appearance) Visualization of Joints and moving parts
Design Cost Overall Design Period
(Time, Trial and Degree of Design Revision
Error) Revision Time(the first, middle and last period)
Expending Time to Revise
KHR-3 Design process
NBh-1 Design Process
Design Process in this research

From the perspective of design quality, which is described upon the dash line in <Table 7>, the pointing
line closer to right side means more efficient the process is. Because it stands for high contribution of design to
robot appearance, and that means process is efficiency. At the side of design cost, pointing lines are showed
under the blue dash line in <Table 7>. In design expenditure, the closer the pointing line comes to the left side,
the more efficient the process is. Because low design expenditure equivalent to high efficiency of process.
Through this assessment we found out that the new design process can effectively provide the design
solution of robot and minimize frequent conflict between the design part and the engineering part.

4. Conclusion
The area of intelligent robot has become clarified. This research was conducted to fulfill social needs,
focusing on the design process and guideline for humanoid robot design. The main procedure were the design
process based on role-character and the concurrent engineering design process, and these approaches proposed a
systematic framework to analyze the design problems of humanoid robots. Especially the concurrent engineering
design process played an important role in integrating the related academic fields in design.
In this research, we set the robot design’s direction and visualized its design. The technical research and
verification of its structure remained for future works to complete the robot design. Realization of robot should
be preceded to verify the structure, and we should conduct the simulation using software and RP(Rapid-
Prototyping) for its realization. We can overhaul the expected problems in structure and elements for its
combination in a short period of time using RP. The refinement of design would be continued through these
approaches.
Recently the diverse research related to robots such as the development of skin tissue for robot and

12
actuator by polymer are being conducted, and as a consequence of that the design expectation about this research
is increasing. As a first step of robot design development, this research was conducted focusing the appearance
design. For future work, the development of robot’s behavior model and interaction design are needed.

Acknowledgements
This work was supported by the Ministry of Information & Communications, Korea, under the Information
Technology Research Center (ITRC) Support Program.

References
[1] Christoph Bartneck, 2004, “A Design-Centred Framework for Social Human-Robot Interaction”, Department
of Industrial Design, Eindhoven University of Technology
[2] Terutsugu Ando, 2005, “Provisional Classification of Service Robots”, JARA
[3] Min-ho Kim, 2003, “A Study for Character Expression Method based on Physiognomy”, Proceedings of the
Korea Society of Design Studies Conference, pp. 98-99
[4] Jose RIVERA-CHANG, “Use of Scenarios as a Collaboration Tool between Industrial Design and
Engineering Students”, Industrial design California State University
[5] Kwan-myung Kim, 1994, “A Study on the Application Concurrent Engineering Concept in Industrial Design Practice”,
KAIST
[6] John R. Hartley, 1992, “Concurrent Engineering” , Productivity Press
[7] Sona Kwak and MyungSuk Kim, 2002, “Emotional Robot Design by Abstracting Emotional Image from
Korean Folk Painting of Tiger”, In proceeding of the 1st Bi-annual Design conference of KSDS, pp.208-209

13