Rev. Téc. Ing. Univ. Zulia. Vol. 39, Nº 5, 101 - 110, 2016
doi:10.21311/001.39.5.15
Gait Design and Simulated Analysis of Quadruped Robot
Lijun Wang, Xiangrui Kong, Shuo Wu, Xiang Wang, Qi Wang, Qingsheng Hu
School of Mechanical Engineering, North China University of Water Resources and Electric Power, Zhengzhou
450045, Henan, China
Abstract
This paper analyses the gait design and simulated analysis based on STC90C516RD + design process. The 12way steering gear is set up in the joint part of the quadruped robot. By changing the duty cycle of the output
PWM signal to control 12 way steering angle and speed, the purpose of controlling various gait of Quadruped
Robot can be achieved. This paper mainly analyzes and designs the two gaits, circular-step and push-ups. This
paper mainly analyzes and designs the two gaits, circular-step and push-ups. The whole design includes the
kinematics analysis, gait analysis and structure design. A three-dimensional model is established base on
SOLIDWORKS and simulation of the model is introduced into the ADAMS. At last the result of simulation is
obtained to verify the design process.
Key words: Quadruped Robot, PWM Signal, Gait Analysis and Design, Simulated Analysis.
1. INTRODUCTION
With the progress of science and technology and the needs of society development, the technique of robot
is extensively used in the product of industry and society life. Industrial robots, for example, are widely used in
manufacturing industries to replace humans with higher risk factors.
Since 1960s, quadruped robots appeared in people's perspective(McGhee,1967). The real first quadruped
robot in the world is made by Frank and McGhee in 1977. This robot has good stability of gait motion, however,
the disadvantage is that the joint of the robot is controlled by a state machine which is composed of a logic
circuit. Therefore, the behavior of the robot is limited, and it only takes the form of a fixed movement. In the
1980s and 1990s, the most representative of the quadruped robots is the TITAN series developed by Japanese
Hirose Shigeo laboratory(Raibert,1990). It has been able to realize the crawling gait, ambling gait, trotting gait,
pacing gait and other symmetrical gait. Contemporaneously, the quadruped robot designed by MITRaibert in
America in 1984 is also very representative. The robot leg adopts telescopic structure, achieves touchdown
buffer and jump by using the cylinder. It can use trot, pace and bound gait to run fast and stablely(Seok, 2014).
The Department of mechanical information science, University of Tokyo, Japan developed a quadruped robot
with intelligent behavior based on vision. The Institute of mechanical engineering, Nan yang Technology
University, Singapore developed a quadruped robot is mainly used in the research of mechanical structure
design, motion control and gait analysis.The development of the quadruped robot starts late in China, but
develops really rapidly. Starting from the beginning of the 1980s, the robot research Institute of Shanghai Jiao
Tong University carry out researches on JTUWM series of Quadruped Robots in 1991. The QW-1 quadruped
robot developed by Tsinghua University has realized the Omni-directional movement under the static gait.
Shandong University(Rong, 2014), National University of Defense Technology, Harbin Institute of Technology,
Beijing Institute of Technology(Li, 2014) and the Shanghai Jiao Tong University developed a number of
hydraulic driven quadruped robots in 2013, which has a substantial increase compared with the previous electric
drive robot in terms of moving speed and weight bearing capacity.
This paper regards the quadruped robot as the example, and mainly talks about the design process of the
quadruped robot which based on STC90C516RD+. Meanwhile, gait design of the two gaits-circular-step and
push-ups are conducted. A three-dimensional model is established base on SOLIDWORKS, simulation of the
model is introduced into the ADAMS and the results of simulation are obtained to verify the design process.
2. CONTROL SYSTEM DESIGN of THE QUADRUPED ROBOT
This design uses STC90C516 microcontroller and Complex Programmable Logic Device (CPLD) to
achieve the production of PWM. Because CPLD has his own parallel processing ability, coupled with a large
number of IO interface, the MCU can control a lot of steering gears at the same time. It can save a lot of space
to prepare for the follow-up work.
But the ability of CPLD can’t handle affairs, so in the actual application, CPLD needs the cooperation of
MCU. Control system of "PC + serial port + Slave computer" is selected. If not, the direct connection of the
101
Rev. Téc. Ing. Univ. Zulia. Vol. 39, Nº 5, 101 - 110, 2016
steering engine by the microcontroller complicates the single-chip processing, but this will make the circuit
simple. The design of the control system structure is shown in Figure 1.
RS232
PC
MAX232
Reset
SCM
CPLD
12-waysteering gear
STC90C516
Crystal oscillator
Infrared
Figure 1. The control system structure
3. THE GENERATION of PWM
The motion of quadruped robot is mainly controlled by the rotation angle and speed rotation of the steering
gear. The control of the steering angle and the rotation speed is controlled by the change of the duty cycle of the
PWM signal.
This design needs the single chip microcomputer to produce a clock cycle for PWM 20ms signal, and use
the interrupt program to adjust the level of the control port. The flow of generating PWM signal is shown in
Figure 2.
Start
Load initial value
of the timer
Waiting for the
interruption
Change counting time
and reserve output pins
Run interrupt programs
Achieved the
maximum number
of interruptions or
not
N
nort
N
End
Y
Figure 2. The flow of generating PWM signal
102
Rev. Téc. Ing. Univ. Zulia. Vol. 39, Nº 5, 101 - 110, 2016
4. GAIT DESIGN of TWO TYPES of QUADRUPED ROBOT
The overall gait design of the quadruped robot includes circular gait and push-up gait (Kavraki L, 1996;
Back S,1999).
4.1. Circular Gait Design
With reference to Figure. 3 we can see that，from the initial position shown in Figure. 3 (a), two legs fold
pairwise and maintain a state of distribution in the positive direction. That is the principal plane of the leg is
perpendicular to the direction of walking shown in figure 3 (b). Then let the quadruped robot clockwise circle,
and get ready for action before making a circle shown in figure 3 (c). Then turn the legs on the diagonal to the
position of the Figure 3(d) in a clockwise direction. And then, another preparation for the next circling is done,
as shown in Figure 3(e). With the reciprocating motion, the robot will be delight in circling and never get tired
of it.
From the turning gait picture, we can found that keep the initial position at the beginning, leg 2&3 and leg
1&4 pairwise folding. In order to make the Quadruped Robot turning clockwise, leg 2&4 need to do a prepare
action before them turn, then turning led 1&3 (Bagchi,1992). Making leg2 close leg3, leg3 close leg2, leg2 close
leg1, leg1 close leg4, makes all the action could be a clockwise entirety. In the next step, turning leg3&1 make
legs to do a prepare action before them turn, then turning led 2&4. Circling gait process is shown in Figure 4.
(a)
(b)
(c)
(d)
(e)
Figure 3. Circling gait
Start
Initialize the timer and
ex-interruption
Send rotation angle and speed to the
designated steering gear
Rotate 1, 3 legs
Rest 2, 4 legs
Y
External
interrupted
Rotate 1, 3 legs
Rest 2, 4 legs
N
Unite 2, 3 legs Unite 1, 4
legs
Rest 1, 3 legs
Rotate 2, 4 legs
Execute external interrupt program
Rest 1, 3 legs
Rotate 2, 4 legs
The steering gear
Reaches the designated position
Y
End
Figure 4. Circling gait process
103
Rev. Téc. Ing. Univ. Zulia. Vol. 39, Nº 5, 101 - 110, 2016
Using Keil writes the program, because in the controller of servo, the range of experiment position is 700 2700. In order to achieve the gait(Bonet B, 2001), the servo position will be add in is shown as Table 1.
Table 1. The servo position of circling gait
Steering
gear
#1
#2
#3
#4
#5
#6
#7
#8
#9
#10
#11
#12
1
1150
1850
1180
1950
1700
1130
1800
1100
1580
1200
1700
1100
2
1150
1350
1180
1450
1700
1530
1800
1500
1580
1200
1700
1100
3
1150
850
1180
950
1700
1530
1800
1500
1580
1200
1700
1100
4
1650
850
1680
950
1700
1330
1800
1300
1580
1200
1700
1100
5
2150
850
2180
950
1700
1130
1800
1100
1580
1200
1700
1100
6
1650
850
1680
950
1500
1130
1400
1100
1580
1200
1700
1100
7
1150
850
1180
950
1170
1130
1800
1100
1580
1250
1700
1100
4.2. Push-up Gait Design
Push-up gait mainly considers the robot can do push-ups like human. But its mechanism is not the same,
the robot has no elbow joint, not like a man in the palm of the ground situation, relies on the biceps and
latissimus dorsi muscle strength. So in such circumstances, I firstly set out the front feet. In order to be more
close to people's way, affixed front foot and the ground can’t touch around, and have to remain motionless state
as push-ups.
From the gait can be seen that at the beginning of the original location, 1, 2 rudders to a certain position to
imitate the human arms; 3, 4 rudders to a certain position to imitate the human legs; 7, 8 rudders lift outside to a
certain position; 11, 12 rudders lift inside to a certain position. It forms a perfect ready posture to push-up.
When 5, 6 rudders lift inside to a certain position concurrently and return to the original position, the
reciprocating motion, constitute a complete push-up position (Fukuoka,2003). Push-up gait process is shown in
figure 5.
Start
Initialize the timer
and ex-interruption
Send rotation angle and speed to
the designated steering gear
Reverse 5, 6 gears
External
interrupted
Return 5, 6 gears
N
Reverse 1, 2 gears
Forward 3, 4 gears
Execute external
interrupt program
Forward 7, 8 gears
Reverse 11, 12 gears
The steering gear
reaches The
designated
position
Y
End
Figure 5. Push-up gait process
Write the program with Keil and the rudder controller on the steering position (the default for the median
1700) range is 700-2700. After actualizing push-up gait motion, adding motion actuator position is shown as
table 2.
104
Rev. Téc. Ing. Univ. Zulia. Vol. 39, Nº 5, 101 - 110, 2016
Table 2. Adding motion actuator position
Steering
gear
1
2
3
4
5
6
7
8
9
10
#1
#2
#3
#4
#5
#6
#7
#8
#9
#10
#11
#12
1500
1500
1500
1500
1500
1500
1500
1500
1500
1500
1300
1300
1300
1300
1300
1300
1300
1300
1300
1300
2040
2040
2040
2040
2040
2040
2040
2040
2040
2040
1000
1000
1000
1000
1000
1000
1000
1000
1000
1000
1600
1500
1400
1300
1200
1200
1300
1400
1500
1600
1330
1430
1530
1630
1730
1730
1630
1530
1430
1330
1100
1100
1100
1100
1100
1100
1100
1100
1100
1100
1800
1800
1800
1800
1800
1800
1800
1800
1800
1800
1380
1330
1280
1230
1180
1180
1230
1280
1330
1380
1500
1550
1600
1650
1700
1700
1650
1600
1550
1500
1500
1500
1500
1500
1500
1500
1500
1500
1500
1500
1300
1300
1300
1300
1300
1300
1300
1300
1300
1300
5. SIMULATED ANALTSIS of THE QUADRUPED ROBOT
In order to verify the feasibility of the whole design and whether the four legged robot can complete the
designed gait smoothly, we take a four legged robot as an example including the simulation and analysis of it.
5.1. Establish a Three-Dimensional Model and Set the Joint Degree of Freedom
Firstly, the three-dimensional model is established by using SOLIDWORKS. Hip joint is between bracket
and thigh where we set up a degree of freedom. And the knee joint is between thigh and leg, we set up another
degree of freedom here. The whole model sets 8 degrees of freedom totally. As shown in figure 6.
Figure 6. Three-dimensional model of the quadruped robot
5.2. Import Model, Add The Drive Function
The 3D model is imported to ADAMS software and constraints are created at the hip and knee joints of
each leg. A total of 8 rotation pairs are created. According to the trajectory of the quadruped robot, we need to
take 8 types of points on the trajectory. The inverse solution is obtained for the angle of the joint. And after
importing the ADAMS to form the Spline curve, the driving function is obtained. Taking the hip and knee joint
of one leg as an example, the image of the two drive functions is obtained. As shown in figure 7 and figure 8.
105
Rev. Téc. Ing. Univ. Zulia. Vol. 39, Nº 5, 101 - 110, 2016
Figure 7. The image of the drive function of Hip joint knee joint
Figure 8. The image of the drive function of knee joint
5.3. Analysis Of The Motion Trajectory
After all the settings are completed, the simulation of a walking process is carried out. Eight moments’
images of each walking process are shown in figure 9 From the whole image we can see that the movement
tracks at the end of the leg are smooth and continuous. No mutation is found. The trajectories of four legs are
similar and only exists a certain phase difference.
Figure 9. Simulation drawing for each time period In the course of a walk
5.4. Analysis of The Joint Angle
The two joints of a same leg (hip joint and knee joint) are considered as the analysis object. Simulation data
of the street corner of the hip joint and knee joint in single leg are shown in Figure 10 and Figure 11. Comparing
the street corner of the joint and the prior input spline curve, we found that they are roughly equal in size, which
means that the input drive spline function meets the requirements of the actual motion.
106
Rev. Téc. Ing. Univ. Zulia. Vol. 39, Nº 5, 101 - 110, 2016
Figure 10. Simulation data of the revolute angle of the hip joint in single leg
Figure 11. Simulation data of the revolute angle of the knee joint in single leg
5.5. Measurement and Analysis of Velocity, Acceleration and Displacement of The Leg
During the process of walking, the robot needs to be continuous and steady. No violent mutation is a
request for the speed, acceleration and leg movement of the thigh, calf, and joint. It requires its slowly and
smoothly change without having a mutation.
Taking a leg (speed of thigh and shank) as an example to illustrate, measurement and analysis of velocity,
acceleration and displacement of the leg are shown in figure 12, figure 13, figure 14, figure 15, figure 16 and
figure 17. It can be seen from the figure that the velocity, acceleration and displacement of the single leg is
relatively flat, and no mutation occurs. The quadruped robot will not fall suddenly. As a consequence, the design
meets the requirements of the actual movements.
Figure 12. Simulation data of the Speed of thigh
107
Rev. Téc. Ing. Univ. Zulia. Vol. 39, Nº 5, 101 - 110, 2016
Figure 13. Simulation data of the Speed of shank
Figure 14. Simulation data of the acceleration of thigh
Figure 15. Simulation data of the acceleration of shank
Figure 16. Simulation data of the displacement of thigh
108
Rev. Téc. Ing. Univ. Zulia. Vol. 39, Nº 5, 101 - 110, 2016
Figure 17. Simulation data of the displacement of shank
5.6. Measurement and Analysis of Displacement of The Bracket Centroid Height
The change of the bracket centroid height is an important basis for stable walking of a quadruped robot.
The simulation is carried out to measure the displacement of the bracket centroid height. As shown in figure 18.
The maximum change of the height of the bracket is not more than 20mm while walking. Its small amplitude
allows it to walk pretty smoothly. As a consequence, the design meets the requirements of the actual
movements.
Figure 18. Simulation data of the displacement of the bracket centroid height
6. CONCLUSIONS
The birth of robots has opened up a new era of human science and technology. As a new tool of human
being, it plays an important role in industrial production and life service. The popularity of robots for improving
productivity is of great significance and great value.
The main content of this design is the control system design of the quadruped robot and the typical gait
analysis. At the same time, a detailed description is made of the motion gait of 12 degrees of freedom robot and
the simulation analysis is carried out to verify the feasibility of this design. The deficiency is that there is the
stability of the robot is slightly worse. Hoping that in the course of future research can be a breakthrough in this
area.
Acknowledgments
This work was supported by Zhengzhou Measuring & Control Technology and Instrumentations Key
Laboratory(121PYFZX181), Innovation Scientists and Technicians Troop Construction Projects of Henan
Province and HASTIT (14HASTIT001), Plan For Scientific Innovation Talent of Henan Province
(164100510018).
REFERENCES
Bonet B, Geffenr H. (2001) "Planning as heuristic search", ArtificiaI Intelligence, 129(12), pp.5-33.
Bagchi A, HatwaI H. (1992) "Fuzzy logic based techniques for motion planning of a robot-manipulator amongst
unknown moving obstacles", Robotica, l0(6), pp.563-573.
109
Rev. Téc. Ing. Univ. Zulia. Vol. 39, Nº 5, 101 - 110, 2016
Back S, Park D-S, Cho J.(1999) "Robot endeffector tracking system based on feedforward neural network",
Robotics and Autonomous Systems, 31(7), pp.43-52.
Fukuoka Y, Kimura H. (2003) "Adaptive running of a quadruped robot on irregular terrain based on biological
concepts", Proc. of IEEE International Conference on Robotics and Automation,14(9), pp.2043-2048
Kavraki L, Svestka P, Latombe J. (1996) "Probabilistic roadmaps for path planning in high-dimensional
configuration spaces", IEEE Transaction on Robotics and Automation, 12(4), pp.566-580.
Li M, Jiang Z, Wang P. (2014) “Control of a quadruped robot with bionic springy legs in trotting gait”, Journal
of Bionic Engineering, 11(2), pp.188-198.
McGhee R B. (1967) “ Finite state control of quadruped locomotion”, Simulation, 9(3), pp.135-140.
Raibert M H. (1990) “ Trotting, pacing and bounding by a quadruped robot”, Journal of biomechanics, 23(Suppl
1), pp. 79-98.
Rong X, Li Y, Meng J. (2014) “Design for Several Hydraulic Parameters of a Quadruped Robot”, Appl. Math,
8(5), pp.2465-2470.
Seok S, Wang A, Chuah M Y. (2014) “Design principles for energy-efficient legged locomotion and
implementation on the MIT Cheetah Robot”, IEEE Transactions on Mechatronics, 20(3), pp. 1117-1129.
110