Research Article

International Journal of Advanced

Robotic Systems
July-August 2017: 1–13
Dynamics and stability analysis on stairs ª The Author(s) 2017
DOI: 10.1177/1729881417720783
climbing of wheel–track mobile robot journals.sagepub.com/home/arx

Xueshan Gao1,2, Dengqi Cui1,2, Wenzeng Guo1, Yu Mu1

and Bin Li2

Abstract
A transformable wheel–track robot with the tail rod whose winding will coordinate the center of gravity of the robot is
researched, and a theoretical basis for the stable climbing of the robot is provided. After a general introduction of the
research, firstly the mechanical hardware and control hardware composition of the wheel–track robot is provided and
the principles of its mechanical structure are illustrated. Secondly, through studying the fundamental constrains during
the process of the robot climbing the obstacles, a mathematical model based on classical mechanics method is built to help
analyze the dynamic principles of a wheel–track mobile robot climbing stairs. Thirdly, the dynamic stability analysis is
carried out by analyzing not only the interaction among forces of track, track edge, and stair step but also the different
stabilities of the robot when the track and the stairs have different touch points. Finally, an experiment of the modeling
track robot climbing the stairs has convinced the effectiveness of the dynamic theories researched, which will be a
beneficial reference for the future mobile robots obstacle climbing studies.

Keywords
Mobile robot, wheel–track, dynamical model, stability analysis, crossing stairs

Date received: 6 March 2017; accepted: 14 June 2017

Topic: Mobile Robots and Multi-Robot Systems

Topic Editor: Nak-Young Chong
Associate Editor: Genci Capi

Introduction
Mobile robots have become increasingly present in human-
related activities either to remove the hazards or to carry large
and critical payloads safely. Mobile robot is capable of moving
around in different environments including stairs and outdoor
grounds. In this circumstances, researches are carried out to
explore extensive applications for mobile robots in special
situations such as rescue, anti-terrorism, removing explosive,
and so on.1,2 In the past decade, both industry and academia
have become more responsive to develop new designs for
mobile robot platforms with better functionality, quality, and
features. At present, there is a focus on designing robots with
the capability of traveling over a variety of surfaces, carrying a
range of payloads, changing into different configurations by
modularity of structure and interface, and fast recovery from
accidental rollover or getting stuck. Today, most of the current
mobile robot platforms still cannot fulfill the said features due
to a fixed structure of its locomotion system.
During the past few years, relevant experts and scholars
have made lots of research on the tracked mobile robot, as a
result many structures of the robot have been generated. For
instance, the “Talon” series of robots from the Foster-Miller
company and the “PackBot” series of robots from the iRobot
1
School of Mechatronical Engineering, Beijing Institute of Technology,
Beijing, China
2
State Key Laboratory of Robotics, Shenyang Institute of Automation
Chinese Academy of Sciences, Shenyang, China
Corresponding author:
Dengqi Cui, School of Mechatronical Engineering, Beijing Institute of
Technology, Beijing 10008, China.
Email: cdq1984@163.com

Creative Commons CC BY: This article is distributed under the terms of the Creative Commons Attribution 4.0 License
(http://www.creativecommons.org/licenses/by/4.0/) which permits any use, reproduction and distribution of the work without
further permission provided the original work is attributed as specified on the SAGE and Open Access pages (https://us.sagepub.com/en-us/nam/
open-access-at-sage).
2 International Journal of Advanced Robotic Systems
Figure 1. Robot mechanical structure. (a) Wheel mode and track mode. (b) System composition.
company, both of which have been applied in the battlefield
to implement various missions, including anti-terror and
anti-explosives, military reconnaissance, and so on.3–5 There
are also certain customized tracked mobile robot for circum-
stances such as disaster relief and fire fighting.6–9 In these
complicated situations, stair climbing has become a basic
function for mobile robots. Other than this, a mobile robot
also needs to be more flexible and is able to climb under
complex terrains. In this case, continuous researches are
required to discover new capabilities of the mobile robot
in the stair-climbing process in different environments.10–12
A new kind of transformable wheel–track robot has been
developed.13–17 To save energy, it walks in wheel mode on
flat grounds and will transfer into track mode on rough road
or stairs as shown in Figure 1: a wheel–track robot only
consists of one simplified and compact coupling recon-
struction mechanism, which enables the robot transfer eas-
ily from walking in wheel mode to track mode, combining
both advantages of a wheel robot and a track robot.
As a result, it is a better choice to have a wheel–track robot
in complex environments such as those with either rough
grounds or changeable stairs. With the flexible capacities, the
wheel–track robot has got a wider application not only for
anti-terrorism and fire rescue but also for anti-terrorism,
removing explosive, and even help the disabled.18–22 Com-
pared to the conventional robot,23–25 the wheel/track robot is
more compact in structure and is more flexible with lighter
weight and smaller volume which consumes less energy.
What’s more, the wheel/track robot applies a coupling recon-
struction mechanism and has a tail rod design that can adjust
the centroid position. The dynamical model and stability anal-
ysis of the robot based on Euler-Lagarange-like systems26–28
are established, which lays the theoretical foundation for the
autonomous control of the robot in the future.
A transformable wheel–track robot that is lighter, more
simplified, and compact has been demonstrated in Figure 1,
which shows a tail rod has been added and by winding the tail
rod will coordinate the center of gravity of the wheel–track
robot to improve the stability of the robot movement. The
mechanism structure and the control system of the robot are
explained in detail in “Robot system” section. The dynamical
model of the robot in stairs climbing has been established and
analyzed through applying the classical mechanics method in
“Dynamic process analysis on climbing stairs” section. A
dynamic stability analysis of the wheel–track mobile robot
is carried out by analyzing not only the interaction among
forces of track, track edge, and stair step but also the different
stabilities of the robot when the track and the stairs have
different touch points in “Dynamic stability analysis of
stair-climbing process” section. The stability of the dynamic
functions is examined in “Simulation analysis and experi-
ment” section, and the examination results have proved the
effectiveness of the theories proposed.
Robot system
Mechanical hardware components
of the robot structure
The robot mainly consists of four parts as shown in Figure
1: the variable structure device, the walking device, the box
body, and the tail rod. The variable structure device is
achieved by a double four-bar linkage mechanism. The
worm gear motors with double output shafts on both sides
of the robot respectively drive the two sets of the main
driving shafts on both left and right sides of the wheel/track
transformation mechanism. The worm gear mechanism is
self-locking, one side of the double output shaft is used to
connect the main driving shaft of the wheel/track transfor-
mation mechanism, and the other side is to install the
potentiometer to complete the angle detection. The walking
wheel device is mainly composed of an internal and
Gao et al. 3

Figure 2. Four-bar linkage variable structure device.

200 40

180
30
Displace (mm)

Velocity (mm/s)
160

140 20

120
10
100

80 0
0 5 10 15 20 2 4 6 8 10 12 14 16
Time (t/s) Time (t/s)
(a) (b)

25
Acceleration (mm/s2)

20

15

10

0
0 5 10 15 20
Time (t/s)
(c)

Figure 3. Displacement, velocity, and acceleration curve of the end point of the swing rod.

external gear ring devices, which is driven by a decelerated
DC motor. In order to make the robot’s structure more
compact, the output torque of the walking motor speed
reducer is transmitted to the gear shaft through a pair of
gears, so that the walking motor on both sides can be
arranged in a staggered way. The two gears of the gear
shaft are separately inner geared with the inner teeth of the
two gears of the walking wheel device; as a result, the inner
and outer gear rings are able to complete synchronous rota-
tion. The tail rod motor can drive one-way wheel rod to
swing to adjust the centroid of the robot.
Take one end point of the swing arm rod of the open
mechanism shown in Figure 2 as the reference point,
carry out kinematics simulation analysis on this point,
and measure the displacement, velocity, and acceleration
curve as shown in Figure 3. It is illustrated in Figure 3
that when the open mechanism operates, the velocity of
the swing rod is slow, and the speed is reduced gently,
which decreases the impact of the elastic track, and this
can satisfy the requirement of the mechanism design.
When the open mechanism opens, the velocity of the
swing rod decreases gently from the maximum speed
4 International Journal of Advanced Robotic Systems
Figure 4. Important mechanical components of the robot.
Figure 5. Control hardware components of the robot.
to zero, and the track is fully open; when it distracts, the
velocity of the swing rod increases slowly to its maxi-
mum speed.
The elastic track shown in Figure 4 will not only deliver
the movement but also stretch or contract frequently to get
adapt to variable working environment. The inner side of
the elastic track has gears that participate in driving mesh
and delivering driving torque, and the outer gears are
mainly used for mobile anti-skidding.
Control hardware components of the robot
The robot control system displayed in Figure 5 adopts a
hierarchical structure, and it is composed of intelligent
plate and control panel which collaboratively complete the
interaction of human and computer, the external environ-
ment detection, and motion control. According to the
change of external environment, the operator will input the
instruction to the intelligent board by the wireless terminal,
and the instruction which will be analyzed in the control
board controls the movements of 2 DC motor and 3 worm
gear motor to realize the control of the robot’s attitude. The
robot’s attitude information is monitored real time by inter-
nal sensors such as gyroscope that are connected on the
control board. This information works not only as the feed-
back references for attitude control but also is transferred
through the intelligent board to the operator which is
reflected in the PC man–machine interface. According to
the feedback information and the real-time external envi-
ronment, the robot attitude requirements are reset and sent
back again to the robot through PC interface or remote
control. In the movement process of the robot, external
sensors including infrared and ultrasonic will send some
external environment information detected to the control
board to assist the operator with some obstacle avoidance
work. The tail rod is driven by the tail rod motor.
Gao et al.
Figure 6. Robot’s climbing stair process.
Dynamic process analysis
on climbing stairs
Process of climbing stairs
When the robot starts to climb the stairs, it will firstly touch
the first step of the stairs. With its variable structure, the
elastic track of the robot will expand when the driving
motors on both sides push the driving heels and the whole
robot forward at the same time step by step during the
climbing process until the entire robot leaves the ground
and keeps climbing up the stairs. This climbing process can
be generally divided into six stages which is shown in
Figure 6: (a) approaching the first step, (b) touch the first
step expanding the elastic track, (c) climb up the first step,
(d) touch the second step totally expanding the elastic track,
(e) climb up the stairs and steady movement on the stairs,
and (f) leave the stairs and steady movement on the ground.
It should be explained that when the robot touch the step, its
back swing rod wheel will touch the ground and when the
robot climb up the stairs, the robot’s bottom track will
climb up onto the step.
Dynamic analysis on climbing stairs of the robot in
tracked mode
The basic structural parameters of the robot are shown as
follows:
R—driving wheel radius,
r—swing wheel radius,
—angle between the bottom track and the ground,
b—width of the step,
h—height of the step,
5
m—robot’s total mass,
v—robot moves forward with constant speed,
O—swing arm wheel’s center as the coordinate origin,
X —axis parallel to direction of bottom track, and
ðxG ; yG Þ—robot’s gravity center in coordinate system.
Under the XOY coordinate system, the distance of the
center of gravity  relative to the origin coordinate whose
angle generated with the X axis is . The two correspond-
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
ing formulas are  ¼ xG2 þ yG2 and  ¼ arctanðyG =xG Þ.
Regarding the point p coordinate that is the contact point
between the track and the stair as ðxp ; yp Þ, under the XOY
coordinate system, the distance of the point p relative
to the origin coordinate k whose angle generated with the
X axis is ’, and the two corresponding formulas are
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
k ¼ xp2 þ yp2 and ’ ¼ arctanðyp =xp Þ.
Figure 6 shows the process that mobile robot climbs the
stairs in the tracked mode. The dynamic model is estab-
lished according to the four phases of the process which are
“Attitude adjudgment before climbing” including Figure
6(b) and (c), “climb up the stair” in Figure 6(d), “steady
movement on the stairs” in Figure 6(e), and “leave the
stairs” in Figure 6(f).
Attitude adjudgment before climbing. Figure 7 describes that
the elastic track of the robot starts to touch the first stair.
The two values Fk1 and NR1 at the right bottom of the
Figure 7 represent the traction generated by the driving
motor and the ground reaction force at the position of the
back swing rod wheel; another two values N1 and f 1 are
separately the supportive force and the frictional forces on
the first step. The angle  is included angle between the
6 International Journal of Advanced Robotic Systems

Figure 7. Robot touches the first step. Figure 8. Robot climbs up the first step.

first step and the track, and  is included angle between

the ground and the track.
In this process, the robot’s dynamic model can be gained
based on the classical mechanics analysis. The following
equation (1) can be concluded by the forces’ equilibrium
equations and its torque equilibrium equation
8
> Fk 1  N 1  sinð þ Þ þ f 1  cosð þ Þ ¼ m€
xG
>
>
>
< NR 1 þ N 1  cosð þ Þ þ f1  sinð þ Þ  mg ¼ m€
yG
 
>
> Q  k sinð’ þ Þ þ r  Wk cosð’ þ Þ  mg cosð þ Þ
>
> Figure 9. Robot ascends on the stairs.
:
m€ xG  ðyG þ rÞ  m€
y  xG ¼ JS  €
G
(1)
In the above equations
Q ¼ f 1  cosð þ Þ  N1  sinð þ Þ;
W ¼ f 1  sinð þ Þ þ N1  cosð þ Þ

The JS in the last equation illustrates the moment of

inertia where the robot moves around the landing spot of
the back swing arm wheel. Also, the frictional force f 1
on the step can be calculated by equation (4), in which
 1 is a frictional factor between the bottom tracks and
the sharp corner of the steps and  1 ¼ 0.1–0.3 is Figure 10. Robot climbs on the stairs.
advisable
f1 ¼ 1  N1 (2) 8 0 0 0 0
>
> F  N 1  sin þ f 1  cos ¼ m€ xG
> k2
>
With equations (1) to (2), the traction Fk in the process < 0 0 0 0
NR 2 þ N 1  cos þ f 1  sin  mg ¼ m€ yG
when the robot touch the first step can be obtained, and > 0 0 0
>
> N 1  h= sin  mg cosð þ  Þ  m€ xG  ðyG þ rÞ
with equation (3), the driving torque Mk can be calculated >
: 0
 m€ yG  xG ¼ JS €
0
Fk ¼ ðMk  M f Þ=R (3) (4)
0
In the equation (3), the total resisting torque M f can Similarly, the traction Fk2 in the process of climbing up
also be known by calculating the caterpillar meshing the first step by equation (4) and the driving torque Mk2 by
frictional force, the resisting torque from the drive equation (3) can both be calculated.
chain, and the frictional force torques of the bearings
on each wheel shaft. Steady movement on the stairs. In Figure 9, after the mobile
Figure 8 provides the illustration of the process that robot leaves the ground and ascends the stairs, it is sup-
the robot is climbing at the first step when the robot’s ported by two or three stairs at the same time, and the
track touches the first step. Equation (4) can be got from supportive force from each step to the robot is Ni;i and
the robot’s forces’ equilibrium equations and its torque Ni;iþ1 . The traction Fk3 and Fk4 are equal to the frictional
equilibrium equation using the classical mechanics force between its bottom tracks and the corners of the stairs.
analysis Figures 9 and 10 illustrate the loading situation of the robot
Gao et al. 7

Figure 12. Force analysis of track and track edge and stair step.

moves exactly to the corner of the top step. g is the angle

formulated between the robot’s bottom tracks and the ground
of the top step, while JN shows the moment of inertia after the
Figure 11. Robot leaves the stairs.
robot twirls around the corner of the top step. The process of
in this climbing process. Equation (5) including forces’ equi- leaving the stairs completes when the robot’s bottom tracks
librium equations and torque equilibrium equation of the touch the ground of the top step, and by this time g is equal to
robot can be got through the classical mechanics analysis zero. Equation (8) below explains the relation between the
8 time variable t, g, and " during this process
> Fk3  mg sinðarctanðh=bÞÞ ¼ 0  
>
> h "
< mg cosðarctanðh=bÞÞ þ N þ N ¼ 0 g ¼ arctan þ ðt  t 0 Þ 2 (8)
11 22
(5) b 2
>
> mg cosð’ þ arctanðh=bÞÞ þ N 11 S
>
: pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
þ N 22 ðS þ h 2 þ b 2 Þ ¼ 0 The equations (7) to (8) provide the way to get the
traction Fk 5 as the robot leave the stairs. Equation (3)
8
> Fk4  mg sinðarctanðh=bÞÞ ¼ 0 provides the driving torque Mk 5 . By referring to the dyna-
>
>
< mg cosðarctanðh=bÞÞ þ N þ N þ N ¼ 0 mical model of the robot in different climbing stages, the
12 23 34
0 (6) change law of the required traction and driving torque can
>
> mg cosð’ þ arctanðh=bÞÞ þ N 12  S þ
>
: 0 p ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi 0 p ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi be summarized.
N23 ðS þ h2 þ b2 Þ þ N34 ðS þ 2 h2 þ b2 Þ ¼ 0
0
In equations (5) and (6), S and S represent the distance Dynamic stability analysis
between the back swing rod wheel and the supportive point of stair-climbing process
on the first step along the slope of stairs. The traction Fk 3
and Fk4 that created during the robot ascends the stairs Interaction between the track and the stair
when the track and the stairs contact in two points can be The force analysis of track, track edge, and stair step is
got by equation (5) and in three points by equation (6). The shown in Figure 12. Ff 1 and Fy1 are the equivalent traction
driving torque Mk3 and Mk 4 can be gained by equation (3). force and equivalent supporting force in coordinate system,
0 0
respectively; F f 1 and F y1 are the actual traction force and
Leaving the stairs. Figure 11 shows the forces in the climbing
actual supporting force, respectively;  is the static friction
process when the tracked robot leaves the stairs and touches
factor of the track and the stair; and  is the angle between
to the ground. Under the action of gravity, the robot twirls
the track and the stair step.
around the sharp corner of the top step until its bottom
00 The necessary and sufficient conditions for the non-
tracks touch the ground. Among these forces, N 1 repre- slipping of the track when the track contacts the stair step
sents the supportive force between the corner of the top step edge line
and the traction. Fk 5 represents the friction force from its
bottom tracks and the corner of the top step. " is the accel- Ff 1
 < < (9)
eration of the robot’s twirling angle. Fy1
Equation (7) including the robot’s forces’ equilibrium
When the track edge contacts the stair step, the necessary
equations and its torque equilibrium equation based on the
and sufficient conditions for firmly hooking stair step are
classical mechanics analysis can be concluded
8 0
F f 1 Ff 1 cos  Fy1 sin
< Fk5  mg sing ¼ m€
> xG   ¼  (10)
00 F 0 y1 Ff 1 sin þ Fy1 cos
N 1  mg cosg ¼ m€ yG (7)
>
:
mgv  ðt  t 0 Þ cosg ¼ JN  " If  < cot

In equation (7), there is a variable t 0 which represents the sin   cos Ff 1 sin þ  cos
  (11)
time when the projective line of the robot’s center of gravity cos þ  sin Fy1 cos   sin
8 International Journal of Advanced Robotic Systems

F
Assuming d ¼ Ffy11 , by equation (11)
8
< sin þ  cos  < cot
>
d max ¼ cos   sin (12)
>
:
þ1   cot

Stability analysis of the robot and the stairs

In the process of climbing the stairs, some assumptions are
as follow: the friction between the driving wheel and the
deformation of the track is not considered, and the move-
ment state of both sides of the track is consistent.
Analysis of one point contact between robot and stair. The
force analysis of contact in one point between robot and
stair is shown in Figure 13. The robot coordinate system
o1 xy is located in the center of the swing arm wheel of
the robot; the total quality of the robot is m; ðx 0 ; y 0 Þ is
the centroid coordinate of the robot in o1 xy coordinate
system; the angle between the robot and the ground is
 1 ; the radius of the drive wheel is R; the radius of the
swing arm wheel is r; the height of the step is h; the
angular velocities of back swing arm wheel, driving
wheel, and front swing arm wheel are o 1 , o2 , and o 3 ,
Figure 13. Force analysis of contact in one point between robot
and stair.
respectively; and the geometric centers of back swing
arm wheel, driving wheel, and front swing arm wheel
are o1 , o2 , and o3 , respectively.
The inertia forces of the direction x and y are

Fax ¼ Gax
(13)
Fay ¼ Gay
ax and ay are the acceleration of the x and y direction,
respectively.
The traction force of the track and ground is composed
of two parts: horizontal traction and vertical traction10

8 8 0 19
>
> < =  
>
> Kj _
s x j
>
> Ffx1 ¼ x N 1 1  exp@ A  cos p þ arctan2ðs_z ; s_x Þ
>
< : maxðjs_x þ ro1 j; jro 1 jÞ ;
8 0 19 (14)
>
> < =  
>
> Kj _
s x j
>
> F z
¼  N 1  exp @ A  sin p þ arctan2ðs_z ; s_x Þ
>
: f 1 z 2
: maxðjs_x þ ro 1 j; jro1 jÞ ;

where x and z are the horizontal and vertical friction where Ff 1 and Ff 2 are the actual traction force of direc-
factor, K the test values for dragging and slipping, and s_x tion x; N1 and N2 the actual traction force of direc-
and s_z the slip velocity of horizontal and vertical direction. tion y; Ff 3 the driving resistance of direction x; G the
When track edge hooks the first step edge, considering track gravity of the robot; and J 1 , J 2 , and J 3 the moment of
slipping of in direction x, no sliding in direction z, the direction inertia of the robot, moment of inertia of driving
z is the vertical direction of the coordinate system o1 xy wheel, and moment of inertia of swing arm wheel,
8 respectively.
>
> s_x ¼ ro 1
< By equation (16)
s_z ¼ 0 (15)
>  
>
: Ff 1 ¼ x N1 1  expðKÞ G cos1 þ Fay  N2
N1 ¼
cos1 þ x expðKÞ sin1
The following equilibrium equations are obtained by " #
Newton Euler and Darren Bell’s principle ðG cos1 þ Fay Þx0  ðG sin1 þ Fax Þðy 0 þ rÞ
8 N 2 ¼ sin 1 =h
> ðFf 1  Ff 3 Þ cos 1 þ N 1 sin 1 þ Ff 2  G sin 1  Fax ¼ 0 € 1 þ J 2 o_2 þ J 3 ðo_1 þ o_3 Þ
þJ1 
>
>
>
>
< ðFf 1  Ff 3 Þ sin 1 þ N 1 cos 1 þ N 2  G cos 1  Fay ¼ 0
> Ff 2 ¼ G sin1 þ Fax  N1 sin 1 þ x N1 expðKÞ cos 1
h
> N2  ðG cos 1 þ Fay Þx 0 þ ðG sin 1 þ Fax Þðy 0 þ rÞ (17)
>
> sin
>
> 1
>
: The stability criterion of the contact in one point
€ 1  J 2 o_ 2  J 3 ðo_ 1 þ o_ 3 Þ ¼ 0
J 1 
between the track and stair step is as follows.
(16)
Gao et al. 9

Figure 14. Force analysis of contact in two points between robot

and stair. Figure 15. Force analysis of contact in three points between
robot and stair.

   
T T T Ff 2
fN1 > 0g fN2 > 0g fFf 2 > 0g 0<  Table 1. Specification of the robot.
N2
(18) Content Unit Symbol Value
Total mass of the robot kg m 26
Analysis of two points contact between robot and stair. When Radius of driving wheel m R 0.175
the center of mass of the robot passes through the first step Radius of two swing arm wheel m r 0.05
Moment of inertia of the robot kg m2 J1 0.69
of the stairs, the robot enters into contact in two points with
Moment of inertia of driving wheel kg m2 J2 0.075
the stairs, and the forces are shown in Figure 14. The fol- Moment of inertia of swing arm wheel kg m2 J3 0.0013
lowing equilibrium equations are obtained by Newton
Euler and Darren Bell’s principle

8
> Ff 1 þ Ff 2  G sin  Fax ¼ 0
>
>
< N 1 þ N2  G cos  Fay ¼ 0
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi (19)
>
> N 2 b2 þ h2  Fay ðy 0 þ rÞ tan þ Fax ðy 0 þ rÞ
>
:
J 1 
€  J 2 o_2  J 3 ðo_1 þ o_3 Þ ¼ 0
where Ff 1 and Ff 2 are the equivalent traction force of
direction x; N1 and N2 the equivalent traction force of
direction y; b the width of the stair; and  the angle between
the track and the stair.
By equation (19)
N1 ¼ N2 þ G cos þ Fay
Fay ðy 0 þ rÞ tan  Fax ðy0 þ rÞ þ J1  € þ J2 o_2 þ J3 ðo_1 þ o_3 Þ
N2 ¼ pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
b2 þ h2

(20)
Assuming Ff 0 ¼ Ff 1 þ Ff 2 ,  0 , 1 , and  2 are the ratio Figure 16. Curve between acceleration and force in one
of the traction force to the support force of the robot in point.
coordinate system.
Two track edges are simultaneously on the stairs The stability criterion of the contact in two points
Ff 0 Ff 1 þ Ff 2 between the track and stair step is as follows
0 ¼ ¼ (21)    
N0 N1 þ N2 T T Ff
fN1 > 0g fN2 > 0g 0  max  d max (23)
Two track edges are not simultaneously on the stairs N
Ff 1 Ff 1 þ Ff 2  N2
1 ¼ ¼
N1 N1
(22) Analysis of three points contact between robot and stair. When
Ff 2 Ff 1 þ Ff 2  N1 there is the contact in three points between the robot and the
2 ¼ ¼
N2 N2 stair step, the forces are shown in Figure 15.
10 International Journal of Advanced Robotic Systems

The following equilibrium equations are obtained by

Newton Euler and Darren Bell’s principle.
8
> Ff 1 þ Ff 2 þ Ff 3  Ff 4  G sin  Fax ¼ 0
>
>
>
> N 1 þ N 2 þ N 3  G cos  Fay ¼ 0
>
>
< N pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
> 2 2
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
2 2
2 b þ h þ 2N 3 b þ h  ðG cos þ Fay Þx 0 þ
> ðFax þ G sinÞðy 0 þ rÞ  J 1 
> €  J 2 o_2  J 3 ðo_1 þ o_3 Þ ¼ 0
>
> pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
>
> N3 b2 þ h2  N 1 b2 þ h2 þ ðG cos þ Fay Þð b2 þ h2  x 0 Þþ
>
>
:
ðFax þ G sinÞðy 0 þ rÞ  J 1  €  J 2 o_2  J 3 ðo_1 þ o_3 Þ ¼ 0
(24)
where Ff 1 , Ff 2 , and Ff 3 are the equivalent traction force of
direction x; N1 , N2 , and N3 the equivalent traction force of
direction y; and Ff 4 the driving resistance of direction x.
It is assumed that the difference between the equivalent
support forces is consistent with the linear distribution law
Figure 17. Curve between acceleration and Ff2/N2 in one point.
N1 ¼ N2  A
1 1
N2 ¼ ðN1 þ N 2 þ N3 Þ ¼ ðG cos þ Fay Þ (25)
3 3
By equation (24)
N3 ¼ N2 þ A

1 ðG cos þ Fay Þx 0
A ¼  ðG cos þ Fay Þ þ pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi þ
2 2 b2 þ h2
(26)
ðG sin þ Fax Þðy 0 þ rÞ tan  Fax ðy 0 þ rÞ þ J 1 
€ þ J 2 o_2 þ J 3 ðo_1 þ o_3 Þ
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
2 b2 þ h2

Assuming Ff 0 ¼ Ff 1 þ Ff 2 þ Ff 3 , 1 , 2 ,  3 , and  4 Simulation analysis and experiment

are the ratios of the traction force to the support force of
the robot in coordinate system. Specification of the robot
Three track edges are simultaneously on the stairs The structural model parameters of wheel–track robot are
shown in the Table 1.
Ff 0 Ff 1 þ Ff 2 þ Ff 2
1 ¼ ¼ (27)
Fy0 N1 þ N2 þ N3

Three track edges are not simultaneously on the Simulation experiment

stairs When a ¼ 0.0001–1.0 m/s2, the relationship curve is shown
in Figure 16 between the acceleration a of the robot and the
Ff 1 Ff 1 þ Ff 2 þ Ff 3  N2  N3
2 ¼ ¼ traction Ff 2 , and the relationship curve is shown in Figure
N1 N1 17 between the acceleration a and Ff 2 =N 2 . By the curves:
Ff 2 Ff 1 þ Ff 2 þ Ff 3  N1  N3 when Ff 2 > 0, N 1 > 0, N 2 > 0, and 0 < Ff 2 =N 2 < 0:5, the
3 ¼ ¼ (28)
N2 N2 requirement of the dynamic stability of the climbing stair
satisfies the dynamic stability criterion of the contact in one
Ff 3 Ff 1 þ Ff 2 þ Ff 3  N1  N2 point between the track and the step of the stairs which is
4 ¼ ¼
N3 N3 provided by the formula (18).
When a > 0.005 m/s 2 , N1 > 0, N2 > 0, and
The stability criterion of the contact in three points 0 < maxð 0 ; 1 ; 2 Þ < 1:832 as shown in Figures 18 and
between the track and stair step is as follows. 19. Based on formula (23), which is the dynamic stability
    criterion of the contact in two points between the track and
T T T Ff
fN 1 > 0g fN 2 > 0g fN 3 > 0g 0  max  d max steps, the condition that the maximum ratio between trac-
N
tion Ff and supporting force N is less than d max satisfies the
(29)
requirement of climbing stair.
Gao et al.
Figure 18. Curve between acceleration and force in two points.
Figure 19. Curve between acceleration and Ff/N in two points.
When a > 0.005 m/s2, N 1 > 0, N 2 > 0, N3 > 0, and
0 < maxð 1 ; 2 ; 3 ;  4 Þ < 1:832 as shown in Figures 20
and 21. Based on formula (29), which is the dynamic sta-
bility criterion of the contact in three points between the
track and steps, the condition that the maximum ratio
between traction Ff and supporting force N is less than
d max satisfies the requirement of climbing stair.
Prototype experiment
When the robot meets the step obstacle, it will switch from
the wheel mode to the track mode, demonstrated in Figure
22(a) and (b). With the robot climbing up the steps, the
inclination angle of the robot become gradually bigger and
there is no sign of leaning forward, then swing the tail rod
to help the robot to change robot attitude as is shown in
Figure 22(d). Under the help of the tail rod, the robot stably
climbs over the following steps and eventually completes
the steps obstacle-surmounting climbing.
11
Figure 20. Curve between acceleration and force in three points.
Figure 21. Curve between acceleration and Ff/N in three points.
The line in Figure 23 represents the driving torque Mk
experiment result that the robot requires for climbing stairs.
The line in Figure 24 shows the experiment result of the
robot’s tilt angle  when the robot climbs stairs. As is
shown, the angle  has a changing process which increases
from zero until it reaches the angle which is the same as
inclined angle of stairs and finally it is back to zero when
the robot leaves the stairs.
Conclusion
Wheel/track coupling type is a new type of mobile com-
bination compared with the traditional wheel/track multi-
ple robots which only contain one set of walking
mechanisms. The structure of a wheel/track coupling type
robot is simple and compact, and it can easily switch
between wheeled mode and tracked mode. As climbing
the stairs is considered as a critical important function to
evaluate the capability of a mobile robot on crossing
obstacles. On the other side, the stability of such a robot
12 International Journal of Advanced Robotic Systems

Figure 22. Robot climbing stairs experiment.

40
to climb the stairs needs also to be ensured. As far as we
know, the achievements gained can provide great refer-
ence for designing and analyzing a mobile robot. Through
Driving torque (N m)

30 researching on the requirements of the robot climbing

stairs, the mechanical structure of a wheel–track mobile
20 robot is designed and the hardware composition of its
control system is explored. After studying the process of
10 the robot climbing the stairs, the dynamics model of the
robot which will provide fundamental support for the sta-
bility analysis can be established.
0
0 0.5 1 1.5 2 2.5 3 3.5 Although the extensive tests and experiments per-
Time (t/s) formed to demonstrate that the robot meets the research
goals, some improvements need to be further addressed in
Figure 23. Relationship of driving torque and time. future developments. The future work is summarized as
follows:

1) Develop more add-on attachments and interfaces for

30 expanding the robot’s functions and applications.
2) Develop an autonomous control system which can be
switched on and off by remote control to assist the
20 user’s operations.
Angle β (°)

Declaration of conflicting interests

10 The author(s) declared no potential conflicts of interest with
respect to the research, authorship, and/or publication of this
article.
0
0 0.5 1 1.5 2 2.5 3 3.5 Funding
Time (t/s) The author(s) disclosed receipt of the following financial support
for the research, authorship, and/or publication of this article: This
Figure 24. Relationship of tilt angle and time. research was supported by State Key Laboratory of Robotics,
Gao et al.
Shenyang Institute of Automation Chinese Academy of Sciences
(grant no. 2015009).
References
1. Davids A. Urban search and rescue robots: from tragedy to
technology. Intell Syst IEEE 2002; 17(2): 81–83.
2. Hebert MH, Thorpe CE and Stentz A. Intelligent unmanned
ground vehicles. Boston, MA: Kluwer Academic Press,
1997, p. 310.
3. Wells P and Deguire D. TALON: a universal unmanned
ground vehicle platform, enabling the mission to be the focus.
In: Proceedings of the SPIE International Conference on
Unmmaned Ground Vehicle Technology VII, vol. 5804, 2
June 2005, pp. 747–757. Orlando, FL: International Society
for Optical Engineer.
4. Birk A and Carpin S. Rescue robotics—a crucial milestone on the
road to autonomous systems. Adv Robot J 2001; 20(5): 595–576.
5. Murphy RR. Human-robot interaction in rescue robotics.
IEEE Trans Syst Man Cybern Part C Appl Reviews. 2004;
34(2): 138–153.
6. Frost T, Norman C, Pratt S, et al. Derived performance
metrics and measurements compared to field experience for
the PackBot. In: Proceedings of the perMIS workshop,
Gaithersburg, MD, USA, 13–15 August 2002, pp. 201–208.
7. Yoneda K, Ota Y and Hirose S. Development of a hi-grip stair
climbing crawer with hysteresis compliant blocks. In: Pro-
ceedings of 4th international conference on climbing and
walking robots (CLAWAR 2001), Karlsruhe, Germany, 24–
26 September 2001, pp. 569–576.
8. Yamauchi B. PackBot: a versatile platform formilitary
robotics. In: Proceedings of SPIE-unmanned ground vehicle
technology VI, vol. 5422, Orlando, FL, 13–15 April 2004,
pp. 228–237.
9. Tao W, Ou Y and Feng H. Research on dynamics and stability
in the stairs-climbing of a tracked mobile robot. Int J Adv
Robot Syst 2012; 9(4): 146.
10. LI N, Ma SG, Li B, et al. An online stair-climbing control
method for a transformable tracked robot. In: IEEE interna-
tional robotics and automation, Minnesota, USA, 14–18 May
2012, pp. 923–929. Minnesota: IEEE.
11. Liu Y and Liu G. Track-stair interaction analysis and online tip-
over prediction for a self-reconfigurable tracked mobile robot
climbing stairs. IEEE Trans Mechatronics 2009; 14(5): 528–538.
12. Rao W, Shi J and Wang J. Analysis of dynamic stability for
articulated—tracked robot climbing stairs. J Mech Eng 2014;
50(15): 60–67.
13. Pector Y, Novoplanski A and Rinberg G. Adaptable traction
system of vehicle: Israel, WO03/029070, April 2003.
13
14. Fu YL, Li HH and Zhang FH. Design and kinematic analysis
of anti-terrorism robot with wheel-tracked variable structure.
Machine Electron 2011; 10: 66–69 (in Chinese).
15. Li J, Bai Y, Jin F, et al. Mechanics analysis of a wheelchair
robot with wheel-track coupling mechanism. J Beijing Inst
Technol 2013; 301–307.
16. Cui D, Gao X and Guo W. Mechanism design and motion
ability analysis for wheel/track mobile robot. Adv Mech Eng
2016; 8(11): 1687814016679763.
17. Guo W. Study on a small special ground mobile robot cou-
pling wheel and track. Beijing, China: Beijing Institute of
Technology, 2016.
18. Novoplanski A, Raizer L and Rinberg G, Traction system:
Israel, WO 2011/004362 A1, January 2011.
19. Wang DX. Mechanical study of barrier-free wheelchair sys-
tem with a wheel-tracked coupling mechanism. Beijing,
China: Beijing Institute of Technology, 2012.
20. Shao-hua KANG, Lin HUANG and Cai-hong S. Design and
analysis on reconfigurable rubber-track variant wheel. Chin J
Construct Machine 2014; 12(4): 320–324.
21. Luo ZR, Shang JZ, Xi L, et al. A Wheel-track complex
tracked robot: China, 201210089271.X, January 2013.
22. Pang AS, Ou YS, Wu XY, et al. Wheel tracked mobile device
and its hub: China, 201310261012.5, October 2013.
23. Comellas M, Pijuan J, Nogu SM, et al. Influence of the trans-
mission configuration of a multiple axle vehicle on the obsta-
cle surmounting capacity. Vehicle Syst Dyn 2014; 52(9):
1191–210.
24. Li L, Wang W, Wu D, et al. Research on obstacle negotiation
capability of tracked robot based on terramechanics. In: 2014
IEEE/ASME international conference on proceedings of the
advanced intelligent mechatronics (AIM), 8–11 July 2014,
pp. 1061–1066. IEEE.
25. Meirion-Griffith G and Spenko M. Simulation and experi-
mental validation of a modified terramechanics model for
small-wheeled vehicles. Int J Vehicle Des 2014; 64(2–4):
153–169.
26. Wang N, Qian C, Sun JC, et al. Adaptive robust finite-time
trajectory tracking control of fully actuated marine surface
vehicles. IEEE Trans Control Syst Technol 2016; 24(4):
1454–1462.
27. Wang N and Er MJ. Direct adaptive fuzzy tracking control of
marine vehicles with fully unknown parametric dynamics and
uncertainties. IEEE Trans Control Syst Technol 2016; 24(5):
1845–1852.
28. Wang N, Er MJ, Sun JC, et al. Adaptive robust online con-
structive fuzzy control of a complex surface vehicle system.
IEEE Trans cybern 2016; 46(7): 1511–1523.