Professional Documents
Culture Documents
Homagni Saha
Bachelor of Technology
in
Mechanical Engineering
by
Homagni Saha
(Roll Number: 112ME0256)
May, 2016
Certificate of Examination
Roll Number: 112ME0256
Name: Homagni Saha
Title of Dissertation: Design and fabrication of biped robot
We the below signed, after checking the dissertation mentioned above and the official
record book (s) of the student, hereby state our approval of the dissertation submitted
in partial fulfillment of the requirements of the degree of Doctor of Philosophy in
Computer Science and Engineering at National Institute of Technology Rourkela. We
are satisfied with the volume, quality, correctness, and originality of the work.
D.R.K Parhi
Principal Supervisor
Department of Computer Science and Engineering
National Institute of Technology Rourkela
Supervisors Certificate
This is to certify that the work presented in the dissertation entitled Design and
fabrication of bipedal robot submitted by Homagni Saha, Roll Number
112ME0256, is a record of original research carried out by him under our
supervision and guidance in partial fulfilment of the requirements of the degree of
Bachelor of Technology in Mechanical Engineering. Neither this dissertation nor
any part of it has been submitted earlier for any degree or diploma to any institute
or university in India or abroad.
Dayal Parhi
Professor
Dedication
This page is dedicated to my parents and my teachers
Signature
Declaration of Originality
I, Homagni Saha, Roll Number 112ME0256 hereby declare that this dissertation entitled
Guidelines for Formatting Dissertation presents my original work carried out as a doctoral
student of NIT Rourkela and, to the best of my knowledge, contains no material
previously published or written by another person, nor any material presented by me for
the award of any degree or diploma of NIT Rourkela or any other institution. Any
contribution made to this research by others, with whom I have worked at NIT Rourkela
or elsewhere, is explicitly acknowledged in the dissertation. Works of other authors cited
in this dissertation have been duly acknowledged under the sections Reference or
Bibliography. I have also submitted my original research records to the scrutiny
committee for evaluation of my dissertation.
I am fully aware that in case of any non-compliance detected in future, the Senate of NIT
Rourkela may withdraw the degree awarded to me on the basis of the present dissertation.
Certificate of Examination ii
Dedication iv
Declaration of Originality v
Acknowledgment vi
Abstract vii
List of Figures x
List of Tables xi
1 Introduction 1
1.1 Problem Definition 1
1.2 Human Robot Interaction 2
1.3 Bipedal and humanoid robots 3
1.4 Why bipedal robots 4
2 Literature review 6
2.1 Brief timeline of development of bipedal robots 6
2.1.1 Preliminary research on waseda legged robots 6
2.1.2 Further developments by the humanoid team at waseda 7
2.1.3 The E series and the P series 7
2.1.4 The humanoid robotics project 9
2.2 Controllers used in bipeds 10
2.3 Sensors used in bipeds 11
2.4 Actuators used in bipeds 11
viii
4 Design and analysis 19
4.1 Overview 19
4.2 Parts of the bipedal robot 22
4.2.1 Joining bracket 22
4.2.2 Contact toe 23
4.2.3 Knee connector 24
4.2.4 Electronics mounting 26
5. Fabrication 27
5.1 Body 27
5.2 Mechatronic systems 28
6. Control 30
7. Navigation 39
8. Results 41
9. Conclusion
References 46
List of Figures
ix
List of Tables
Table 4.1 Comparison of the joints and their range of angular operation angles 16
Table 6.2 Gait parameters of the Biped Robot 25
Table 6.3 DH Parameters for the Right Leg chain of the Biped Robot 29
Table 7.1 Input member functions 35
Table 7.2 Output member functions 35
x
Chapter 1
Introduction
1
1.2 Human robot interaction
Bipedal robots were designed with the sole purpose of mimicking humans in the form of
humanoid robots. Hence it can be reasoned that biped robots are a part of the research
on humanoid robots, thus to justify the research knowledge must be gained as to why
research on humanoid robots are essential. Again for that it is essential to understand the
role of human robot interaction in the modern society, and that humanoid robots are the
ultimate manifestation of the functionalities exhibited by the class of robots built for the
cause of human robot interaction. In general human-robot interaction, also known as HRI,
is an emerging field which studies the interaction between humans and robots. The
advances made in the areas of artificial intelligence have made robots more and more
capable of doing human-like tasks. This has increased the importance of making robots
more user-friendly and operable in the human environment.
Robots are fake operators with abilities of discernment and activity in the physical world
frequently alluded by scientists as workspace. Their utilization had been limited to
industrial plants, however these days they have a tendency to be found in the most
technologically advanced social orders such as in rescue missions, military fight, mine
and bomb location, logical investigation, law authorization, entertainment and hospitality.
In this context, bipedal robots are also being increasingly used in military testing sites in
order to simulate the effects of various parameters on soldiers movement. These new
spaces of utilizations mean a nearer association with the client. The idea of closeness is
to be taken in its full significance, robots and people share the workspace additionally
share objectives as far as accomplishment of an assignment. This nearby association
needs new hypothetical models, on one hand for the mechanical autonomy researchers
who work to enhance the robots utility and then again to assess the dangers and
advantages of this new "companion" for our cutting edge society. If a robot needs to
operate in a home, homes have a great deal more mind boggling social principles than
those in processing plants or even military situations. In this way, the robot needs seeing
and comprehension abilities to fabricate dynamic models of its environment. It needs to
sort questions, perceive and find people and further their feelings. The requirement for
element limits pushes forward each sub-field of mechanical autonomy.
2
1.3 Bipedal and humanoid robots
A humanoid robot is a robot with its body shape worked to look like that of the human
body. A humanoid configuration may be used for practical purposes, for example,
associating with human apparatuses and situations, for exploratory purposes, for
example, the investigation of bipedal walking, or for many other different purposes.
Humanoid robots have a middle torso, a head, two arms, and two legs; however some
types of humanoid robots may display just part of the body, for instance, from the waist
up. Some humanoid robots likewise have goes to imitate human facial components, for
example, eyes and mouths. Androids are humanoid robots worked to stylishly take after
people. Bipedal robots focus mainly on the legs and are made to concentrate on bipedal
locomotion.
Tasks involving dexterous manipulation, efficient perception of the environment,
localization in unknown areas and moving around in cluttered spaces are more efficiently
handled by robots having human like morphology. Bipedal motion is one of the important
aspects of this morphology and it is unique among all other types of locomotion, in that it
can be made the most energy efficient and also it has the most flexibility in motion.
Wheels definitely have greater speed, but are unable to traverse gaps, steps, and uneven
terrain, thus legged robots have their own advantage. In most cases a bipedal locomotion
system consists of a system of linkages actuated by a series of actuators which can be
either rotary or linear. The degrees of freedom are fixed by the design and it also affects
the manner of walking. The controller implements the control algorithm for the
coordinated motion of the actuators which determine the various parameters of walking,
which are also known as the gait parameters.
3
Figure 1.1: Basic anatomy of bipedal robot showing specific interest points [1]
4
supplanting connections or gathering of connections by point masses, and so forth. The
improvements are done relying upon specific necessities, contingent upon how the issue
is engaged concerning case inactive elements, limit cycle strolling, zero minute point
control, among others. These rearrangements are frequently reached out to recreations
where the arrangement regularly works splendidly on the grounds. The disentanglements
amid re- enactments can be go significantly facilitate by presenting changes in the
topology of the first framework. In spite of the fact that the just specified rearrangements
have demonstrated viability in various applications, when all is said in done there is no
criteria to apply them. A large portion of them depend on the theory that the mass is
aggregated at the focal point of mass of the framework or subsystem being concentrated,
yet no further examination is done regardless of whether this suspicion is legitimate. From
this short dialog we can presume that defended disentanglements will be invited to
enhance the current situation with the present art.
5
Chapter 2
Literature Review
6
2.1.2 Further developments by the humanoid team at waseda
WAM-1 was first developed in 1967, in which artificial muscles made of rubber
indigenously developed by WASEDA were used. WAM-1 had seven degrees of freedom
(DOF), three in the arm and four in the hand. Later after several versions of
improvements WAM-4 was developed which could detect objects using tactile and visual
sensors bound to its fingers. These were followed by the introduction of the musician
robot WABOT-2 which could accompany a person in music and play music on
keyboard[4].In 1995 the humanoid robotics project of the WASEDA University produces
the HADALY and WABIAN (WAseda BIpedal humANoid), in order to realize and evaluate
close exchanges of information with humans. The manipulator system of HADALY was
developed by sugano laboratory which introduced a passive compliance adjustment
mechanism, also known as the MIA(Mechanical Impedance Adjuster) in all the seven
joints in the hand and the arm subsystem[5]. The entire design of WABIAN was complete
until February 2004 and human like walking was achieved with knee stretch posture.
Later the Wabian 2R released in 2006 , had verisimilitude to three dimensional walking
due to incorporation of flexible pelvis system along with stretched knees and had the
unique flexible toes with the curving arch, allowing it to lift off at its toe after landing on its
heels first[6]
In 1986 the Honda Motor Company started its long journey towards the development of a
fully-fledged humanoid robot, by creating its first collection of bipedal robots known as the
E-series. E0 to E6 were completed by 1993 followed by the creation of the p series[6].E0
had the lowest degrees of freedom and walked the slowest completing each cycle in 5
seconds. Parallely in 1985, Hitachi Ltd. Released their biped robot WHL-11 which could
walk on flat surface and turn, but took 13 seconds between each step, while E0 took 5
seconds. Progressive improvement in walking speed and flexibility was observed after
increasing the degrees of freedom, development of dynamic movement and optimization
of dimensions.E6 was able to walk over obstacles, balance itself and climb stairs [7].
Scientists observed the basic difference between slow walking and fast walking where a
smoother variation of the position of centre of gravity with changing position of sole of foot
was necessary to achieve higher walking speeds.
7
Figure 2.1: Left side shows slow walking and right side shows fast walking [7]
8
Figure 2.2: Dynamics of walking [8]
Among a few lacunas of this impressive humanoid one which is well known is that the
robot is unable to sit down and stand up, as well as stand up from a lying position on
ground. Both these features are easily achieved in NAO and HRP-2, as discussed
later.[9]. In spite of these minor defects ASIMO remains a yardstick of comparison for the
performance of humanoid robots even today. The latest version of ASIMO released in
2011, with 57 degrees of freedom is quiet capable to be a personal assistant. Be it
smoothly walking up the stairs or pouring liquid from bottle to the glass, performing sign
language or even jumping on one foot the delicacy of control is a wonder even today.[10]
The Humanoid Robotics Project (HRP) project, also known as promet, was put forward
to bring about the development of general level domestic tasks oriented helper robots. It
was sponsored by METI (Ministry of Economy, Trade and Industry, Japan) and (NEDO)
New Energy and Industrial Technology Development Organization; with the Kawada
Industries as the leading initiator. At the starting Honda P3 robot was bought and was
transformed into the 1600 mm tall 120 kg heavy HRP-1 imbued with 30 DOF and with
features such as the telecommand [11].Later increased focus was given on movement
and flexibility and the HRP-2P was developed using the experiences from the advanced
leg module HRP-2L.The unique feature of HRP-2P was that it had a cantilever type hip
joint structure and a waist with two degrees of freedom.
Its performance was upgraded by using cooling system in its leg actuators and
redesigning of feet for a high stiff structure [12]. One major advantage of the cantilever
9
structure joint was that it could prevent tipping over by changing the shape of the support
polygon. [12]As for the waist joint a smoother gait could be achieved because the
moment generated in the yaw axis could be suppressed using waist motion, which can be
seen as compensation applied for high speed walks.Hirohisa Hirukawa,Shuuji Kajita,
Fumio Kanehiro, Kenji Kaneko and Takakatsu Isozumi in their research suggested a
walking pattern generator based on a hybrid approach taking into account both the ZMP
and the inverted pendulum model. The HRP-3P was developed in the phase two of the
humanoid robotics project. Stress was laid on designing components so as to provide
protection against dust and water, also a distributed control system was employed. It has
a cantilever hip joint and a 2 dof waist joint. As a strong reason for using the waist joint
the developers aver to the fact that waist motion can suppress yaw axis moment
generation while walking [13].The HRP series was further modified into the latest versions
of the androids HRP 4 and HRP-4C, which is a feminine looking humanoid robot,
standing 158 centimetres tall and weighing 43 kilos along with a battery pack with a
lifelike head, face and figure of a young Japanese female in the average standards. There
were major leaps taken in the area of design for lightweight and slim body like
miniaturization of drive system, distributed control system and developing tiny and
distributed motor drivers.[14]
10
2.3 Sensors used in bipedal robots
All humanoids use cameras for impression of nature and article acknowledgment. They
also, contain sensors which are widely utilized as a part of different parts, either to
interface with clients or the situations. Touch sensors are utilized especially as a part of
the hands, feet and middle. Another sensor is an inertial estimation unit (IMU), which
incorporates an accelerometer, gyrator and, some of the time, a magnetometer in one
joined framework. Sound tactile frameworks are utilized for voice communications. GPS
and separation sensors (ultrasonic or laser based) encourage route inside and outside.
execution of the robot and consideration must be taken in selecting both the incitation
frameworks and movement transmission extras. Some extra plan components are strain
rate, vitality thickness (power/volume), particular force (power/weight), effectiveness,
electromechanical coupling, lifecycle, power utilization and sources, repeatability and
reaction time. The development of the innovation, simplicity of control and frill request
further thought.
12
Chapter 3
Bipedal locomotion
models
3.1 Inverted pendulum model
Small modifications to the dynamics of the classical pendulum will give rise to the inverted
pendulum model which is a common model used to describe bipedal locomotion. One of
the major advantages of this model is that it provides us with a complete analytical
solution, because of its dynamic equations which are simple. On the other side, there is a
flaw in the manner in which the model is constructed. The mass of the robot is
represented by the mass of the pendulum with the location being at the centre of mass of
the robot. The pendulum rod is pivoted at the point which is the systems centre of
pressure. This leads to ignoring a few other important dynamic effects. First of all, the
basic free body diagram of an inverted pendulum is considered as shown in the diagram,
in figure 3.1.
Thus eliminating the rate of change of angular momentum term using 3.2 and 3.3, we get,
The two equilibrium points of the pendulum are at the angles where theta values are 0
and 90 degrees. At 0 degree, however there is unstable equilibrium and at 90 degrees
there exists stable equilibrium. The motivation for this model is drawn from the fact that
humans support themselves on a single leg during major portion of the walking cycle. The
centre of pressure concept for the toe contact with ground is essential for application of
this model. An example for locating the centre of pressure in the human toe is shown in
figure 3.2 below.
14
by stopping more than one foot and feel how the response strengths of the floor changes
when the balance is re-established. Critical rotational body flow is lost in the past models;
as was said, for straightforwardness some penances are required. A standout amongst
the most alluring qualities in the past models is the capacity to improve the mind boggling
contact strengths in one point (the focal point of weight). So as to keep the first one point
contact, and in the meantime, add the ability to apply torque over its backing. The mass
can pivot its focal point of mass and it is upheld by a massless leg. Contingent upon the
setting of work, two models emerge, the one utilized on 2D cases called the response
wheel pendulum and the 3D case called the response mass pendulum. Two separate
pendulums are shown in figure 3.3.
Figure 3.3: Equivalence of reaction wheel pendulum [31]. a)Inverted pendulum fixed to
the floor and moved by applying torque against floor b) If to the flywheel torque is applied
to accelerate it, both the the models are equivalent
In 3.3(a), a rotational joint bounds the inverted pendulum to the ground. Torque can be
applied by the joint to move the pendulum in the desired manner. The system is
described by the equation
The separate pendulum in 3.3(b) is unable to apply torque directly to ground; it can cause
acceleration of a flywheel, so its mathematical description includes equation (3.5) along
with equation 3.6 shown below
Where I denotes the inertia of the mass and the term multiplied with it is the flywheels
absolute angle. The inertia moment in 2D is a genuine number, however on account of
3D movement a tensor should be utilized. This tensor is ascertained around the focal
point of mass of the biped, and it is made by all the inflexible connections that are a piece
of it. The tensor of inactivity can likewise be connected with its equinomental ellipsoid to
build the altered pendulum.
15
Figure 3.4: Reaction mass pendulum [31]
In Figure 3.4 a representation of the response mass pendulum can be seen. On account
of an inflexible body the ellipsoid of inertia is altered. At the point when the same ellipsoid
is ascertained for a multi body framework the centroidal inertia moment of the complete
framework changes its quality relying upon the setup; in this manner, changing its
response mass pendulum as indicated by the condition of the framework. The last, opens
the likelihood to discover new control systems by modelling the inertia points. This is
better shown in Figure 3.5. The model talked about in the present segment adds
imperative changes with a specific end goal to keep complex flow of the framework,
however in the meantime it is entirely straightforward.
Figure 3.5: Inertia shaping method [31]. The inertia ellipsoids form is changed based on
the orientation of the robot.
that not a sensible one, and to augment the vitality pumped into the framework is the one
appeared in Figure 3.6. The pendulum begins at 1, then when it is situated in the most
minimal part of the cycle 2 it is raised to 3. At the point when the pendulum achieves 4 it
has a more extensive swinging abundance, after that it is taken to 5. There are a few
applications for biped robots; the most evident is to control the swinging of the free leg
amid the single position stage. This has been examined as a contrasting option to
recuperate vitality on aloof element strolling [32]. They utilize an under activated robot
with no motors at the hip, the robot has a cylinder that lift the leg when swinging. The
robot and its step cycle are appeared in Figure 3.7. Another understood wonder is the
determined pendulum [36]. At the point when a pendulum is energized under certain
recurrence it has a stable upward position as appeared in Figure 3.8(a). As was seen
already in this section, the biped dependability issue can be outlined to the security of a
rearranged pendulum; this is utilized as a part of [37] to plan the stride of a compass
biped robot with extensible legs. A schematic of this robot and its walk cycle can be found
in Figure 3.8(b). The cycle is fundamentally the same as the one appeared in Figure 3.6,
yet for this situation it is connected to the position leg rather than the swinging one.
Despite the fact that on reproductions the framework functions admirably, there are no
reports of an effective working biped. The conceivable purposes behind this issue depend
on the move from point 4 to point 5. To accomplish the move the robot would need to pull
the floor, yet as a general rule the position leg can just push against it.
17
Figure 3.7: Parametric excited robot and stick diagram showing its gait [34]
Figure 3.8: Driven pendulum model a) Driven pendulum phenomenon schematic figure b)
Parametric excitation optimal trajectory for a bipedal robot. [35]
At long last, it ought to be noticed that there are broad answers for the parametric
constrained pendulum. The conditions analysed in this area just study what is occurring in
a solitary stride. The explanation behind this choice is that a simply parametric wonder is
unrealistic to accomplish on a biped robot in light of the low frequencies present in
ordinary walk.
18
Chapter 4
Modelling and analysis
4.1 Overview
A major issue in the design of bipeds is the designing of legs; significant research has
already been done in this field. With the aim of realizing fast humanlike walking motion,
Lohmeier et al. [38] focused on the design of leg for Lola i.e. a walking humanoid robot
having 22 DOF. The robot is light in weight and boasts a joint design having many
numbers of sensors employing brushless motors. Knee and ankle joints were driven by
linear actuators for a good mass distribution in legs. Some designers focus on
methodologies that forge together passive and active walking dynamic models for a
suitable gait [39].
To mimic the human bipedal walk for an elaborated study on the same a bipedal robot
has been designed and fabricated with the help of PLA which is a 3D printing plastic
material for light and durable use. The fabricated biped has a total of 8 Degrees of
Freedom (DOF), having 4 DOF of in each leg such as Hip Pitch, Hip Yaw, Knee Pitch and
Ankle Roll. For each DOF, a servo motor (HiTEC HS-985MG) of 5 volt input is used.
19
RHY DOF 4
LHY DOF 5
BASE
RHP DOF 3
LHP DOF 6
RKP DOF 2
LKP DOF 7
RAR DOF 1
LAR DOF 8
YAW
PITCH
Figure 4.1: Skeleton model of the Biped Robot with all DOF
The skeleton model of the biped has been made in solidworks to facilitate study of
kinematics. In the model all the rotations of the biped are clearly mentioned as shown in
Figure 4.1. The cylinders represent the axes of rotation of each revolute joint and reveal
individual degrees of freedom For ankles two roll rotations are there as Right Ankle Roll
(RAR) and Left Ankle Roll (LAR), for knee two pitch rotations are there as Right Knee
Pitch (RKP) and Left Knee Pitch (LKP), and for hip two yaw with two pitch rotations are
there as Right Hip Yaw (RHY), Left Hip Yaw (LHY), Right Hip Pitch (RHP), Left Hip Pitch
(LHP).. The kinematic model is made up of seven links which approximate the locomotion
features analogous to those of lower part of human body.
The waist consists of a rectangular channel that houses the power unit and the controller.
This is supported by the framework of the two legs. The joint connecting the waist to the
legs has two degrees of freedom in the transverse plane and is in the sagittal plane. This
joint is followed by the shin which ends in the knee joint having its freedom in the sagittal
plane. The degrees of freedom have been modified in the end effector torso from that in
the sagittal plane, which is most commonly implemented to that in the coronal plane. The
unique walking pattern is developed and simulated in SolidWorks according to this
modification. A stable closed cycle walking pattern has been successfully implemented
and the biped is able to turn around obstacles. The unique S shaped jointed structure of
the leg enhances static and dynamic stability and effectively houses the servo motors.
20
Comparison and perfect decision on the range of joint values for the controlling servo
motors is a crucial factor in the design of bipeds. Table 1 illustrates the comparison as
well as the joint ranges for the biped.
Table 4.1: Comparison of the joints and their range of angular operation angles
The easy walking of robot synchronised motor rotation is very important aspect. For that,
a good algorithm has been developed in ARDUINO language and for the control of biped
an ARDUINO controller board is implemented in the biped. ARDUINO Mega 2560 has
been used here which has an input voltage of 5 volt and a good memory capacity for the
program. Good obstacle avoidance has been achieved by the feedback of two ultrasonic
sensors implemented at the hip of the biped. Figure 4.2 shows the front view of the
working biped model.
21
Figure 4.2: Front view photo of the Biped robot
22
Figure 4.3: Joining bracket
Since the joining bracket is made up of plastic material, it is susceptible to failure by
breakage by simply tearing off or shearing due to the high torque of the servo motor.
Hence a stress analysis of the joining bracket is carried out using the FEA package within
SOLIDWORKS 2012. The results are shown in figure 4.4.
This is a small part in the knee region of the bipedal robot connecting the two servo
motors that provide ankle pitch and hip pitch respectively. Its main role is to appropriately
24
increase the length of the foot so that desirable gait parameter is obtained. It also
provides some clearance between the two pitching motors. The CAD model of this part is
shown in figure 4.7
The deformation shown in the image has been scaled to 23. The von misses stress lie
within the safe zone for the material selected. Hence the design is safe.
25
4.2.4 Electronics mounting
This is a sort of accessories part, but has its own contribution to the design, although it
does not support the robot mechanical, not performs a specific function related to gait
synthesis, it is important in holding the various electronic parts of the robot as a whole
together with the model. It consists of various fixtures for fixing the controllers and
sensors and grove like arrangements for proper passage of jumper wires. These are fixed
to the chest and the body of the robot using strong adhesives, and these are also 3D
printed.
26
Chapter 5
Fabrication
5.1 Body
Excepting the mechatronic segments, the biped was created completely by 3D printing
utilizing the Prusa I3 Rework 3D printer utilizing PLA material. Polylactic Acid (PLA) is a
thermoplastic material with a yield stress of 57.8 MPa .There are a few favorable
circumstances in utilizing 3D printers, for example, the capacity to make complex shapes
and for lessening producing time, cost and the measure of squandered material. For this
situation, producing the bipeds perplexing and mind boggling parts, for example, the knee
joint, servo section holder and contact toes, would be extremely troublesome without 3D
printing. Be that as it may, there are additionally a few impediments of 3D printing utilizing
a FDM (intertwined testimony displaying) machine. One of the restricting components of
the Prusa I3 Rework 3D printer is the span of the base printable element, which is around
0.178 mm (0.007"), the size for the thickness of one layer of PLA plastic. This confined
the measure of point of interest in the model. The surface roughness of 3D printed parts
additionally assumes a noteworthy part in moving parts, creating grating and confining
movement. In the 3D printing process, if there is insufficient resilience, the parts will meld,
rendering them pointless. A few tests were done to describe the impact of the hole
between moving parts. It was found that leaving a separation of 0.45 mm between moving
parts would minimize the erosion and permit the parts to move (for parts imprinted in get
together mode). Another burden of the assembling procedure in the present 3D printers is
that PLA plastic is fundamentally debilitated when 3D printed. In the event that those
parts could be 3D printed with a much more grounded material, the issue would be
disposed of. Supplanting the maximally-stacked thermoplastic PLA areas of biped with
metal materials will keep up joint portability and kill disappointments because of feeble
plastic.
3D printer used: PRUSA I3 REWORK 210x210x200
Material used: PLA
Software used: Repetier Host version 1.0.6
Resolution: 0.0125
Accuracy: 0.1mm
The relative extent of manufacture materials and printing time of the bipedal robot are
shown in Figure 5.1.
27
Figure 5.1: Relative extent of manufacturing materials and printing time.
This figure demonstrates the required model material, bolster materials (conciliatory
materials that are broken up by solvents after the parts are created) and the required time
of manufacture for every part of the robot. It can be seen that the aggregate time of
printing the robot is ~300 h. The figure is critical, since it let us know the essential data in
3D printing biped robots relying upon the size and the capacity of the robot. The 3D
printer gives the precise measure of backing and model materials when a CAD record is
stacked to its working programming. Along these lines, the extents in Figure 5.1 are
precise qualities without a mistake bar.
28
Figure 5.2: Specifications of servo motor
As for the controller, ATMEGA 2560 was used which runs in the ARDUINO interface. All
the codes for gait synthesis were written in the ARDUINO interface. In order to sense
distances, a ping type ultrasonic sensor is used and in order to maintain balance an
accelerometer is used.
29
Chapter 6
Control
6.1 Locomotion variables
Before we analyse the biped kinematically, all the different variables which influence the
walking gait has to be considered implicitly when programming the controller for stable
walking gait. All such parameters are described in Figure 6.1.
hr
(x3, y3, z3)
l1 l3
l2 l4
sl
Figure 6.1: Diagram describing locomotion parameters
Ps: Time Period of the walking cycle
t: time
S: The step length
hr: Hip ripple height
H: Hip height
fc: Maximum foot clearance
sl: Stance length
l1: shin length of right leg
l2: foot length of left leg
l3 : shin length of left leg
30
l4: foot length of right leg
l5 : Hip length
(x1, y1, z1): Centre coordinates of Ankle
(x2, y2, z2): Centre coordinates of Knee
(x3, y3, z3): Centre coordinates of Hip
The values of the above parameters have been found out for the biped using practical
experiments and verified with simulation results from Solidworks. The values obtained are
shown in Table 2.
Table 6.2: Gait parameters of the Biped Robot
Parameters Link Value
l1 = l3 115 mm
Legs l2 = l4 125 mm
l5 280mm
S 88 mm
hr 16 mm
H 318 mm
Walking Parameters
fc 16 mm
sl 75 mm
Ps 15 ec
31
Each domain encompasses a set of reference trajectories and its own control strategy. A
brief description of each stage is given below:
1. Pushing off: At time T=T1, the biped is at zero position and in the double support
phase, the swing leg is lifted off by pushing the ground.
2. Single support stage: At time T=T2, the swing foot completely loses contact with
the ground. Now the knee of the swing leg is locked by swinging the leg forward.
3. Striking: This time at T=T3, the knee of the swing leg is completely locked. Now it
moves forward to strike the ground.
4. Double support: After striking the ground at T=T4, double support phase is
reached again. The stance leg and swing leg simultaneously swap their positions.
This process is so fast and spontaneous in humans that stage 2 seems to appear
right after stage 4. But in the biped the control is closed loop and the biped returns
to stage 1 noticeably, with a pause after stance leg completes its function [41].
Several researchers have often placed emphasis on controlling CM and ground contact
forces, as it is observed during each walking cycle CM oscillates over the supporting limb
like an inverted pendulum.
The Linear inverted pendulum model is improvised upon by employing the ZMP criterion
in designing as mentioned in [42].The ZMP in this approach is kept fixed during stepping
motion in the middle of the supporting foot sole, enhancing stability while the centre of
mass follows the path shown by linear inverted pendulum. The relative position of the CM
and the ZMP within the support polygon is also crucial in controlling balance during slow
walking [43]. Figure 6.3 shows the stick model of the biped robot walking in various
stages.
l5
CM
ZMP ZMP
Unstable Position Stable Position
Figure 6.3: Relative Position of Centre of mass (CM) and ZMP in the support polygon
32
6.3 Kinematic Analysis
As biped has multiple degrees of freedom, the kinematic analysis is necessary to define
the parameters of the kinematic chain which provides the transformation from joint space
to Cartesian space that paves the way to dynamic analysis. Again from the dynamic
analysis the forces and torques required in actuators can be calculated. The Quaternion
space and the Cartesian space are the two primary systems of spaces used in kinematic
analysis of different part of robots. One Cartesian space can be reached from another
using a simple translation and rotation. In such cases, 4x4 homogeneous transformation
matrix is used most often.
Although there exist several techniques to represent rotation like Cayley-Klein, Gibbs
vector parameters, orthonormal matrices, Pauli matrices, axis and angle, Hamiltons
quaternions, and finally Euler angles; in this paper Euler convention is used. Instead of
solving the six independent equations obtained from homogenous transformation
matrices one can use the Denavit & Hartenberg [44] parameters consisting of only four
variables to fully define the transformation from one Cartesian system to another. The
orientation of any three-dimensional body needs nine elements to fully define its position
using homogeneous transformations, on the other hand, the system of dual quaternions
requires only four. Using the kinematic parameters one can predict the required
trajectories of motion of certain parts to achieve desired motion, as was done in various
researches on NAO robots to write precise programs for kicking the ball in Robosoccer.
Let F be the transformation matrix, Rx Ry Rz be the basic rotation matrices, and L be the
translation matrix. The base position or the starting coordinate of the biped is taken at the
centre of the rectangular channel. The numbers in parenthesis signify the x, y and z
values of the translation matrix respectively.
Thus, the analysis is started from the base point of the biped to end effector i.e. hip to toe
through some rotations and parameters as described in Table 3. At each step while
changing the coordinate system from one to another, values of the four DH parameters
are to be substituted in the equation of FDH matrix.
Table 6.3: DH Parameters for the Right Leg chain of the Biped Robot
Frame (Joint) a d
The zero position of the biped is fixed at the top part of biped as shown in Figure 6.4
which would lay the reference coordinate system.
34
O
Figure 6.4: AutoCAD model of the Biped Robot showing the base point
The analysis has been carried out from one reference frame joint i.e. base to the
reference point of another joint using the transformation matrix F DH which is realized by
multiplying the two rotation matrices and two translation homogeneous matrices given by
the DH parameters.
FDH Rx Lx a Rz Lz d ..............................(1)
The equation 1 gives the following analytical form of the resulting matrix
cos sin 0 a
sin cos cos cos sin d sin
FDH ..............................(2)
sin cos cos sin cos d cos
0 0 0 1
After substituting the DH parameters total eight matrices has been derived among which,
one matrix is the translation from base point to right hip yaw joint, next five matrices are
transformation matrices from base to end effector and the rest two are rotation matrices to
align the coordinates of end effector with the global coordinate system. Multiplying all the
eight matrices the resultant transformation matrix W is obtained to reach the end effector
starting from the base point. All the calculation to obtain the equation is done by using the
software MATLAB due to the complexity.
The final transformation matrix for the right leg can be calculated by the equation 3.
End
FBase L0Base F0RHY FRHY
RHP RKP RAR End
FRHP FRKP FRAR R y R z ( )..............................(3)
2
35
The similar conditions has been applied to get the transformation for the left leg where the
only change is Base point that is (-40, 55, -14); while the translation, transformation and
rotations remain similar to right leg.
The actual numeric matrix is a dynamically changing matrix with all its elements
continuously changing during the walking cycle of the biped but following a fixed pattern
depending on the gait. The joint variables are expressed in terms of these varying
elements in the matrix and the constant elements like the length of the links in the biped.
The joint space angles are defined in the sagittal, frontal and transverse planes as shown
in Figure 6.5.
7
3
1 5
8
4
36
Equating the two matrices W and N we get a non-linear system of equations below.
N11
sin 2 cos 3 sin 3 cos 2 .........................(5)
cos 1
N 21
sin 2 cos 3 sin 3 cos 2 ..........................(6)
sin 1
From the above two equations, value of 1 is calculated.
N 21
1 tan 1 ........................................................(7)
N11
To solve these equations, an additional geometric constraint is imposed. The right leg of
the biped in the sagittal plane can be shown during the walking cycle as in Figure 6.6.
2
l2
D
3
3'
1
l1
l12 l22 D 2
3' cos 1 ...........................................(8)
2l1l2
D x1 x3 z1 z3
2 2
............................(9)
Where,
Again,
3 3'.......................................................(10)
By substituting the values of 1 and 3 in equations (4) and (5) 2 is calculated as:
1 l12 l22 D2
2 cos
2l1l2
2 2
sin N11 N21 ................................(11)
1
cos 4 sin 3 cos 2 sin 1 sin 2 sin 1 cos 2 cos 3 N33 ................(13)
37
N N
tan 4 32 4 tan 1 32 ................................(14)
N 33 N 33
In the same manner, values of 5, 6, 7 and 8 for the left leg can be calculated.
sl 2 t 2 t
x1 t sin .............................(15)
2 Ps Ps
fc 2 t
z1 t 1 sin ...................................(16)
2 Ps
S 2 t 2 t
x3 t sin .............................(17)
2 Ps Ps
z3 t H ..............................................................(18)
y2 sl sin Ps t ........................................(19)
And
The equations (15) to (18) give us the trajectories in the sagittal plane while equation (19)
gives the same for knee in the frontal plane.
38
Chapter 7
Navigation
Fuzzy interface development for obstacle avoidance
Fuzzy logic is widely utilized as a part of portable robot route. In fuzzy logic, fuzzy logic
controllers incorporate distinctive heuristics control instruments in various structure e.g.
assuming then, else-if and so on guidelines which manufacture a proficient robot having
distinctive humanly like subjective and quantitative usefulness with control stream
components. Controller ought to be as keen as human or in a better way to stay away
from hinders which happens while performing exercises to reach to the destination. There
might be diverse sorts of impediments on the way and subsequently it ought to give an
effective system to classify the deterrents and make a move in like manner. Fuzzy logic
permits to numerate calculation and to coordinate typical thinking in a characteristic
structure of the framework. There are diverse stream controls in fuzzy logic control
framework and is appeared in the fig-7.1.The initial phase in this control procedure is the
fuzzy controller will contemplate the surroundings, and this is to be evaluated by various
infrared sensors. After this the data will be passing on to the fuzzy logic control framework
through the data extraction framework. Presently
The prime choice must be taken by the fuzzy logic controller for snag shirking and way
mapping. This data must be gone to the robot utilizing certain sort of circuit. The robot will
take after a specific way as indicated by the data given by the fuzzy logic controller.
Presently it's the ideal opportunity for impediment evasion, the movement of the robot is
totally recognition based activity, on the off chance that he saw any hindrance in its way
then he need to take after a crash freeway which is chosen by the fuzzy logic controller.
Presently this must be taken after till the robot ranges to its destination. The complete
stream graph is shown in fig-7.1.
As per the data gained by the sensors, the responsive practices are chosen by the fuzzy
logic controller framework/fuzzy logic calculation to choose the turning speed of the biped
robot. The proposed fuzzy logic framework comprises of four segments: fuzzification,
fuzzy standard, fuzzy interface, and defuzzification. The inputs for the fuzzifier are the
data picked up by the sensors; now this data is likewise contribution to the fuzzy set
characterized in; the characterized fuzzy set is described by various fuzzy part work like
close, medium, far, left_obs, right_obs, front_obs, head_ang and the control parameters
moderate, med, and quick. These part capacity are utilized to fuzzify the speed of the
robot turning speed as left_turn and right_turn individually, this specific thing is embraced
39
from [4].
The input parameters for the fuzzy controller are illustrated in table 4.
Table 7.1: Input member functions
Fuzzy set Member functions Parameters
Left Obstacle 2 Trapezoidal Near , Far
1 Triangular Medium
Right Obstacle 2 Trapezoidal Near, Far
1 Triangular Medium
Front Obstacle 2 Trapezoidal Near, Far
1 Triangular Medium
Heading Angle 2 Trapezoidal Positive , Negative
1 Triangular Straight
The output member functions of the fuzzy controller are summarized in table 5
40
Chapter 8
Results
8.1 Gait generation
From Table 3 for dimensional and gait parameters of the biped robot, we substitute all the
obtained locomotion variables in the equations (15) to (18) and solve the equations using
MATLAB to get the following graph which shows the trajectories of Hip and Ankle in x and
z coordinates for both legs.
X Trajectory of ankle and hip for 3 cycles of walking Bibed Z Trajectory of ankle for 3 cycles of walking Bibed
3000 60
x trajectory of left Ankle z trajectory of left Ankle
x trajectory of right Ankle z trajectory of right Ankle
2500 x trajectory of Hip 50
Z Position in milimeter
x Position in milimeter
2000 40
1500 30
1000 20
500 10
0 0
0 5 10 15 20 25 30 35 40 45 0 5 10 15 20 25 30 35 40 45
Time in Second Time in Second
41
Figure 8.2: Yaw Pitch and Roll of the tip of the ankle in a walking cycle
In figure 10 it can be observed that at the end of the stage 3 only the rate of change of roll
in the ankle tip is negative, and in fact the roll is the controlled parameter in the ankle.
This signifies the striking stage of the left ankle with the ground which is why a negative
rate of change is essential to maintain stability.
Figure 8.3: Yaw Pitch and Roll of the Hip in a walking cycle
Figure 8.4: Yaw Pitch and Roll of the right knee in a walking cycle
Control algorithm used for controlling the servo motors plays a very important factor in
dynamic analysis which is a future application of this research. A robust control is
achieved as a dynamic similarity between programmed and implemented angular
displacements is observed. The Figures 8.5 and 8.6 gives a brief insight into the control
algorithm used in the biped walking and shows the variation of the joint motor angle at
42
each individual joint within the walking cycle during walking.
(a) (b)
Figure 8.5: Joint Angles at different joints of both the legs in a walking cycle (a) at Ankle
(b) at Knee
Figure 8.6: Joint angles at Hip joint of both the legs in a walking cycle
8.2 Navigation
VREP is simulation software best suited for single as well as multi robot walking with
versatility in sensor implementation. It follows the programming language LUA, which is
based on ANSI C language. VREP is chosen here for the simulation of humanoid due to
its properties like collision detection, minimum distance calculation and better path or
motion planning. By taking the PSO algorithm and the fitness function explained above, a
program has been written in LUA language and simulation of the NAO humanoid robot
has been carried out in VREP software and a path has been obtained as explained in the
figure 8.7 below [6].
43
Figure 8.7: Simulation result of biped robot using fuzzy controller
44
Chapter 9
Conclusion
In this paper, a new approach to the biped model is described which proposes 4 DOF at
each leg and a total of 8 DOF. Available literatures on biped leg design have been
surveyed and unique configuration for the joints have been conceived. The forward and
reverse kinematics of the new biped is described using DH convention. The program
written in ARDUINO has given a good result in stable walking of the biped having a good
walking cycle. A gait trajectory has been calculated with some equations where periodic
gait functions are considered for obtaining the leg trajectories. The rotation angle and joint
trajectory can now be predicted easily. The walking of the biped is simulated in
Solidworks software, where the suitable rotation angle and joint torques are validated for
stable walk, and good results have been observed.
Biped motion is an exploration field with heaps of unanswered inquiries, how people
combine walking styles is an unsolved matter. Albeit noteworthy accomplishments have
been done in the field, new apparatuses to plan or to investigate these frameworks are
constantly invited. The idea of element equality in this theory has been confined for the
2D case as it were. Albeit legitimate for the majority of the cases, since strolling can be
constantly decoupled as two tasks, 3D changes can be important when contemplating
complex circumstances. Naming some of them could recuperate from expansive
irritations, or a moving robot performing some pirouettes. In any case, it ought to be seen
that even in complex circumstances the legs can be still be displayed utilizing 2D models
since they are thin components if they are not totally collapsed.
45
References
[1] Kwon, SangJoo, and Jinhee Park. "Kinesiology-Based Robot Foot Design for
Human-Like Walking." International Journal of Advanced Robotic Systems 9 (2012).
[2] Biped Walking Robot.http://www.humanoid.waseda.ac.jp/booklet/kato_4. html
[3] Takanishi, A., et al. "Realization of dynamic biped walking stabilized with trunk motion
under known external force." Advanced Robotics: 1989. Springer Berlin Heidelberg,
1989. 299-310.
[4] Sugano, Shigeki, and Ichiro Kato. "WABOT-2: Autonomous robot with dexterous
finger-arm--Finger-arm coordination control in keyboard performance." Robotics and
Automation. Proceedings. 1987 IEEE International Conference on. Vol. 4. IEEE, 1987.
[5] Hashimoto, Shuji, et al. "Humanoid robots in Waseda universityHadaly-2 and
WABIAN." Autonomous Robots 12.1 (2002): 25-38.
[6] Ogura, Yu, et al. "Human-like walking with knee stretched, heel-contact and toe-off
motion by a humanoid robot." Intelligent Robots and Systems, 2006 IEEE/RSJ
International Conference on. IEEE, 2006.
[7] http://world.honda.com/ASIMO/history/e0/
[8] Obringer, Lee Ann, and J. Strickland. "How ASIMO Works." HowStuffWorks, Inc 2005
(1998): 28.
[9] https://en.wikipedia.org/wiki/Humanoid_Robotics_Project
[10] Sakagami, Yoshiaki, et al. "The intelligent ASIMO: System overview and integration."
Intelligent Robots and Systems, 2002. IEEE/RSJ International Conference on. Vol. 3.
IEEE, 2002.
[11] Hirukawa, Hirohisa, et al. "Humanoid robotics platforms developed in HRP."Robotics
and Autonomous Systems 48.4 (2004): 165-175.
[12] Kaneko, Kenji, et al. "Design of prototype humanoid robotics platform for
HRP."Intelligent Robots and Systems, 2002. IEEE/RSJ International Conference on.
Vol. 3. IEEE, 2002.
[13] Akachi, Kazuhiko, et al. "Development of humanoid robot HRP-3P." Humanoid
Robots, 2005 5th IEEE-RAS International Conference on. IEEE, 2005.
[14] Kaneko, Kenji, et al. "Humanoid robot hrp-4-humanoid robotics platform with
lightweight and slim body." Intelligent Robots and Systems (IROS), 2011 IEEE/RSJ
International Conference on. IEEE, 2011.
[15] Byko, M. Personification: The Materials Science and Engineering of Humanoid
Robots. Available online:http://www.tms.org/pubs/journals/JOM/0311/Byko-0311.html
(accessed on 20 October 2015).
[16] Kajita, S.; Kaneko, K.; Kaneiro, F.; Harada, K.; Morisawa, M.; Nakaoka, S.I.; Miura, K.;
46
Fujiwara, K.; Neo, E.S.; Hara, I. Cybernetic human HRP-4c: A humanoid robot with
human-like proportions. In Robotics Research; Springer: Berlin, Germany; Heidelberg,
Germany, 2011; pp. 301314.
[17] Park, I.W.; Kim, J.Y.; Lee, J.; Oh, J.H. Mechanical design of the humanoid robot
platform, HUBO. Adv. Robot. 2007, 21, 13051322. [CrossRef]
[18] Lee, D.D.; Yi, S.J.; McGill, S.; Zhang, Y.; Behnke, S.; Missura, M.; Schulz, H.; Hong,
D.; Han, J.; Hopkins, M. Robocup 2011 humanoid league winners. In Robocup 2011:
Robot Soccer World Cup XV; Springer: Berlin, Germany; Heidelberg, Germany, 2012;
pp. 3750.
[19] Tsagarakis, N.G.; Metta, G.; Sandini, G.; Vernon, D.; Beira, R.; Becchi, F.; Righetti, L.;
Santos-Victor, J.; Ijspeert, A.J.; Carrozza, M.C.; et al. Icub: The design and realization
of an open humanoid platform for cognitive and neuroscience research. Adv. Robot.
2007, 21, 11511175. [CrossRef]
[20] Cheng, G.; Hyon, S.H.; Morimoto, J.; Ude, A.; Hale, J.G.; Colvin, G.; Scroggin, W.;
Jacobsen, S.C. Cb: A humanoid research platform for exploring neuroscience. Adv.
Robot. 2007, 21, 10971114. [CrossRef]
[21] Hara, Y. Sonys Qrio Robot Hits the Ground Running. Available online:
http://www.eetimes.com/document.asp?doc_id= 1147759 (accessed on 20 October
2015).
[22] Schwarz, M.; Pastrana, J.; Allgeuer, P.; Schreiber, M.; Schueller, S.; Missura, M.;
Behnke, S. Humanoid teensize open platform nimbro-op. In Proceedings of the 17th
RoboCup International Symposium, Eindhoven, The Netherlands, 30 June 2013.
[23] AsimoTechnicalInformation.Availableonline:
http://asimo.honda.com/downloads/pdf/asimo-technicalinformation.pdf (accessed on
20 October 2015).
[24] Kajita, S.; Kaneko, K.; Kaneiro, F.; Harada, K.; Morisawa, M.; Nakaoka, S.I.; Miura, K.;
Fujiwara, K.; Neo, E.S.; Hara, I. Cybernetic human HRP-4c: A humanoid robot with
human-like proportions. In Robotics Research; Springer: Berlin, Germany; Heidelberg,
Germany, 2011; pp. 301314.
[25] Park, I.W.; Kim, J.Y.; Lee, J.; Oh, J.H. Mechanical design of the humanoid robot
platform, HUBO. Adv. Robot. 2007, 21, 13051322. [CrossRef]
[26] Tsagarakis, N.G.; Metta, G.; Sandini, G.; Vernon, D.; Beira, R.; Becchi, F.; Righetti, L.;
Santos-Victor, J.; Ijspeert, A.J.; Carrozza, M.C.; et al. Icub: The design and realization
of an open humanoid platform for cognitive and neuroscience research. Adv. Robot.
2007, 21, 11511175. [CrossRef]
[27] Cheng, G.; Hyon, S.H.; Morimoto, J.; Ude, A.; Hale, J.G.; Colvin, G.; Scroggin, W.;
Jacobsen, S.C. Cb: A humanoid research platform for exploring neuroscience. Adv.
Robot. 2007, 21, 10971114. [CrossRef]
[28] Geppert, L. Qrio: The Robot That Could. Available online:
47
http://spectrum.ieee.org/robotics/robotics-software/qrio-the-robot-that-could (accessed
on 20 October 2015).
[29] Dzahir, M.A.M.; Yamamoto, S.I. Recent trends in lower-limb robotic rehabilitation
orthosis: Control scheme and strategy for pneumatic muscle actuated gait trainers.
Robotics 2014, 3, 120148. [CrossRef]
[30] Tadesse, Y. Actuation technologies for humanoid robots with facial expressions
(hrwfe). Trans. Control Mech. Syst. 2013, 2, 110.
[31] Lucero, Alba, and David Mauricio. "Kinematic and dynamic analysis for biped robots
design." (2012).
[32] At L. Hof. The extrapolated center of mass concept suggests a simple control of
balance in walking. Human movement science, 27(1):11225, February 2008.
[33] Arthur D. Kuo, J. Maxwell Donelan, and Andy Ruina. Energetic consequences of
walking like an inverted pendulum: step-to-step transitions. Exercise and sport
sciences reviews, 33(2):8897, April 2005.
[34] Fumihiko Asano, Zhi-Wei. Luo, and Sang Hyon. Parametric excitation mechanisms for
dynamic bipedal walking. In Robotics and Automation, 2005. ICRA 2005. Proceedings
of the 2005 IEEE International Conference on, pages 609615, 2005.
[35] Toyoyuki Honjo, Z.W. Luo, and Akinori Nagano. Parametric excitation of a biped robot
as an inverted pendulum. In IEEE/RSJ International Conference on Intel- ligent
Robots and Systems, 2008. IROS 2008, pages 34083413, 2008.
[36] David Sudor and Steben Bishop. Inverted dynamics of a tilted parametric pendulum.
European Journal of Mechanics, 18(3):517526, 1999.
[37] Toyoyuki Honjo, Z.W. Luo, and Akinori Nagano. Parametric excitation of a biped robot
as an inverted pendulum. In IEEE/RSJ International Conference on Intelligent Robots
and Systems, 2008. IROS 2008, pages 34083413, 2008.
[38] Lohmeier, S.; Buschmann, T.; Schwienbacher, M.; Ulbrich, H.; Pfeiffer, F., "Leg
Design for a Humanoid Walking Robot," Humanoid Robots, 2006 6th IEEE-RAS
International Conference on, pp.536-541, Dec. 2006.
[39] Roa, Mximo, Diego Garzn, and Ricardo Ramrez. Design Methodology for Biped
Robots: Applications in Robotics and Prosthetics. INTECH Open Access Publisher,
2007.
[40] Van Zutven, Pieter, Dragan Kosti, and Henk Nijmeijer. "On the stability of bipedal
walking." In Simulation, Modeling, and Programming for Autonomous Robots, pp.
521-532. Springer Berlin Heidelbe.
[41] Lacquaniti, F., Y. P. Ivanenko, and M. Zago. "Kinematic control of walking." Archives
italiennes de biologie 140, no. 4 (2002): 263-272
[42] Kajita, S., Kahehiro, F., Kaneko, K., Fujiwara, K., Harada, K., Yokoi, K., Hirukawa, H.,
Biped Walking Pattern Generation using Preview Control of the Zero Moment-Point,
Proceedings of IEEE International Conference on Robotics and Automation, vol.2, pp:
48
1620 - 1626, May 2003.
[43] Tang, Zhe, Zengqi Sun, Changjiu Zhou, and Lingyun Hu. "Reference trajectory
generation for 3-dimensional walking of a humanoid robot." Tsinghua Science &
Technology 12, no. 5 (2007): 577-584.
[44] J. Denavit, and R. S. Hartenberg, 1955, A kinematic notation for lower pair
mechanisms based on matrices, Trans ASME Journal of Applied Mechanisms, Vol.
23, pp. 21522.
[45] Cuevas, Erik, Daniel Zaldivar, Marco Perez, and Marte Ramirez. "Polynomial
trajectory algorithm for a biped robot." arXiv preprint arXiv:1405.5937 (2014).
[46] Sorao, Kenji, Toshiyuki Murakami, and Kohei Ohnishi. "A unified approach to ZMP
and gravity center control in biped dynamic stable walking." In Advanced Intelligent
Mechatronics' 97. Final Program and Abstracts., IEEE/ASME International
Conference on, p. 112. IEEE, 1997.
[47] Asano, Fumihiko, and Zhi-Wei Luo. "Asymptotically stable biped gait generation
based on stability principle of rimless wheel." Robotica 27, no. 06 (2009): 949-958.
[48] Jacob, SanjeevPothen, and Krushna Shankar Sethi. Navigation of Mobile Robot using
Fuzzy Logic. Diss. 2013.
49
50