Proceedings of The 2016 IEEE International

Conference on Real-time Computing and Robotics

June 6-9, 2016, Angkor Wat, Cambodia

Design and Analysis of a Two-DOF Coupling Motion Robotic Joint

He Huang, Erbao Dong*, Member, IEEE, Lin Zhou, Zhuo Duan, Chunshan Liu, Jin Luo, Min Xu, Jie Yang

Abstract— In this paper, a robotic joint with a two-DOF these counterbalance mechanisms. The robot manipulators
coupling motion is proposed for higher payload ability, power are bulky and the movability are usually limited due to the
density and speed. With these advantages the joint is ideal for counterbalance mechanism. Several new kind of two-DOF
use in robots, especially multi-DOF manipulators. The entire
robot manipulator composed of these joints is not involved in joints were proposed for the high compact structure [9]-[11].
this paper. It will be reported in detail in a future publication. Robot manipulators with these joints can cut off the number
Here we just focus on the robotic joint. This robotic joint which of the joints for reducing own weight and volume. But in
is configured as a series-parallel hybrid structure can be called comparison to the manipulators composed of one-DOF joints
as angular swivel joint. And the joint can move on a spherical the payload ability and power density are still the same due
surface whose apex angle is 100◦ by the coupling motions
of two independent motor-reducer drive systems. Details of to their serial structures.
the mechanism design, kinematics analysis and differential In the purpose of comprehensive improvement the per-
kinematics will be depicted. In addition, all motions of the formance of robot manipulators, we present a two-DOF
joint can be divided into two modes: mode 1 is that the two coupling motion robotic joint. This joint consists of two
motor-reducer systems work in the same speed and with the concentric kinematic paths, the external passive universal
opposite direction (SSOD); and mode 2 is that the two motor-
reducer drive systems work in the same speed and with the same joint (U-joint) which only permits yaw and pitch motions
direction (SSSD). Moreover, a unique control strategy based on between adjacent links and an internal angular swivel joint
these two basic motion modes are presented. Finally, simulations [12-13] which provides the two-DOF motion: deflecting and
and experiments are implemented to verify the feasibility and orienting. The presented joint can effectively increase the
performance of the joint. payload ability, power density and execution speed. Note as
I. INTRODUCTION increase in this paper means that the joint can realize higher
performance than other joints under the same driving system.
In the past several decades, robot manipulators have expe-
This concept was first applied in a snake robot for a more
rienced a vigorous development in many fields [1]-[3]. The
compact design by Takanashi [14-15]. However the spherical
development can’t be realized without the technology ad-
design of the angular swivel joint was so cumbersome.
vancement. Plenty of research works focused on mechanism
Based on this design a new two-DOF joint mechanism was
design, path planning, control strategy, etc. which aims at
proposed to mimic the humanlike motion in [16-17]. But the
improving the performance of the manipulators have been
two driving systems were put on the same side that leads to
done. But some problems including low payload ability, low
the whole joint so bulky and the backlash was introduced.
power density and slow execution speed still exist and limit
In [18] a three-DOF joint was proposed for spatial hyper-
the development of robot manipulators.
redundant robots. This three-DOF joint not only allowed
Researchers have attempted many methods to improve
for a compact and high stiff design but made the structure
the situation. Although these methods above had worked in
more complicated. And the power density was reduced by the
some aspects, there still no appropriate solution for solving
distal degree of freedom. In [19] researchers put the U-joint
all these problems. For example, in the case of industrial
into the angular swivel joint and the two drive systems on
manipulators [4-5] the payload to own weight ratio are
both sides but instead of concentrically. The results was that
usually more than 10:1. A lot of the manipulators power
the extra transmission mechanism was introduced as well as
is used for supporting their own weight instead of payload.
the unbalanced force.
The result is that the high-performance motors and high
In this paper, we make huge improvements of the mech-
reduction ratio speed reducers are used to provide sufficient
anism. Firstly, we extend the scope of the workspace by
power to support self-weight and payload. In [6]-[8] multi-
increasing the angle of the inclined planes. Secondly, a U-
DOF counterbalance mechanisms were suggested to with-
joint is installed outside and the shape of angular swivel
stand their own weight. And these methods really worked.
joint is changed from spherical to cylinder for high stiffness
However a multi-DOF manipulator always need several of
and compact. So we can put the motor shaft concentric
*This work was supported by National Natural Science Found of China with the rotation axis of the cylinder. By this way, the
(No. 51275501. 51105349 and 6137095) and the Fundamental Research unbalanced forces acted on the two cylinders are eliminated,
Funds for the Central Universities.
He Huang, Erbao Dong*, Member, IEEE, Lin Zhou, Zhuo Duan, Chun- respectively. Moreover, we propose a new kind of coupling
shan Liu, Min Xu, Jie Yang are with Department of Precision Machinery control strategy which has a great difference from other joints
and Precision Instrumentation. University of Science and Technology of using decoupling control strategy [20]-[22]. By this way we
China, Hefei, Anhui Province, 230026, China. Jin Luo is with Hefei-
Eagle Automation Engineering Technology Co., Ltd(corresponding author to can finally improve the comprehensive performance of the
provide phone:+86-0551-63601482; e-mail:[email protected]) joint.

978-1-4673-8959-4/16/$31.00 © 2016 IEEE 573

 The reminder of the paper is organized as follows. In with respect to U-joint. So the orienting of two cylinders
section II, mechanism design of the robotic joint, the joint changes as well as the normal n. Then stop joint 1, move
kinematics, differential kinematics and motion modes are joint 3. The result is that point A moves in another circle
provided in detail. Section III explains the control strategy 2. Circle 1 and 2 are the same except the center position
using the combination of two different motion modes and of circle 2 which moves with the normal vector n. After
experimental results are presented to validate the modeling repeating the above processes the whole workspace of joint
and performance of the robotic joint. Finally, the conclusion can be obtained. In Fig. 2(b) the shaded area is the workspace
and future work are discussed in section IV. of the joint which is a spherical crown.
II. MECHANISM SYSTEM DESCRIPTION
A. Detailed mechanical description
This joint shown in Fig. 1 consists of two dependent
mechanism systems: one passive U-joint system and the
two motor-reducer-cylinder systems (in the reminder of this
paper, the motor-reducer-cylinder system is simply referred
to as a drive and all the angles will be mentioned below are
divided by the reduction ratio of the speed reducer). The U-
joint is designed placing outside the cylinders as the output.
The internal drive systems act as the actuators and skeleton.
The U-joint can only permit yaw and pitch motions for the
roll motion is limited by the two cylinders. The motion along
the axes of the two drives is limited by the U-joint likewise.
Due to this characteristic the robotic joint is easy sealing by
the bellows and the electrical wires can also be put inside the
joint or along the outside surface of the joint without been
torsional stressed to damage.
Fig. 2: The process of explaining the motion principle and obtaining the
workspace of the joint from the geometric structure: (a) only joint 3 is
rotated from the initial position and (b) the workspace of our joint.

B. Kinematics analysis
The presented robotic joint with two-DOF coupling mo-
tion is analyzed by forward and inverse kinematics. Since
the joint is a strong coupling system. To solve the forward
kinematics, firstly, we need to decouple the joint into two
separated parts as mentioned before, the angular swivel joint
(two drive systems and two cylinders with inclined planes)
and the U-joint. Secondly, solving the rotation matrix of these
two parts. Finally, the forward kinematics of the robotic joint
can be solved by combining the two equal rotation matrix.
Since we have a spherical workspace, we can use spherical
coordinates to depict the output of the joint. The angles α
Fig. 1: Design of the robotic joint: (a) the initial position of the joint and and β denote the deflecting and orienting angles which are
(b) the joint is in a deflecting position. defined in Fig. 3.
The model of the internal part U-joint is shown in Fig.
Geometrically, the motion principle and the workspace of 4. As mentioned before, the U-joint which is never twisting
the joint can be explained in Fig. 2. The angular swivel joint around its axis but only deflecting by the composite motion
consists of three joints: joint 1, 2 and 3. The U-joint can be of the yoke 2 and cross 0 when yoke 1 is fixed on base 1.
separated as two joints: joint 4 and 5. Joint 1 and 3 are the And the output of joint is fixed together with yoke 2. The
active joints and joint 2, 4 and 5 are the passive joints. As transformation matrix of U-joint can be easily expressed as
only joint 3 moves in the initial position, point A moves in
a circle 1 shown in Fig. 2(a) whose center is on the normal
vector n of the inclined plane. As joint 1 and 3 move at the R35 = Rot(y3 , θ1 )Rot(x4 , θ2 ) (1)
same speed and direction in the initial position. Since the U-
joint limits the roll motion of the whole joint, it is clear that And the relationship between spherical coordinates and
the U-joint cant move but the two cylinders move as a whole U-joint coordinates can be written as

574
 Fig. 5: Coordinate system of the drive system, (a) is the initial position of
the drive systems and (b) is the position after the drive systems rotated by
Fig. 3: The definition of the deflecting angle α and orientating angle β . φ1 and φ2 respectively.

frames a, b, 0 and 2 are coincident with each other and the

frames a and b are coincident with each other. Frame 0
is the reference frame attached to base 1 and frame 2 is
the output frame attached to base 2. Both of them are same
with the frames defined in the coordinates of the U-joint.
Frames a and a are attached to cylinder 1. The relationship
between frames a and a is a rotation ψ (ψ is the angle of
the inclined plane) about the xa axis. Frames b and b are
attached to cylinder 2. The relationship between frames b and
b is a rotation ψ about the xb axis. When the two drives
are rotated by φ1 and φ2 separately. The relative rotational
angle between the two inclined planes is θ . We can obtain
the relationship of frame 2 with respect to frame 0, which is
written in Eq. 3.

 
R02 = R0a Raa Rab Rbb Rb2
(3)
= R(Z0 , φ1 )R(Xa , ψ)R(Za , θ )R(Xb , ψ)T R(Zb , φ2 )T
The two transforms, should be equal for this joint. Thus the
forward and inverse kinematics can be obtained by equaling
the correspondent values of the two transforms which is
Fig. 4: Coordinate system of the U-joint.
written in Eq. 4.

sψ sφ1 cθ1 − cψ sθ1 = sψ sφ2

sψ cφ2 cθ2 + cφ sθ2 = sφ1 sψ
α = cos−1 (cθ1 cθ2 ) (4)
cφ1 sθ2 + sφ1 sθ1 cθ2 = c12 wψ
−sθ2
β = sin−1 (  ) cφ2 cθ1 cθ2 + sφ2 sθ1 = −c12 wψ
c2θ2 s2θ1 + s2θ2 (2)
where cx = cos(x) and sx = sin(x) for x = φ1 , φ2 or ψ. And
θ1 = tan−1 (tα cβ ) c12 = (1 − cθ1 cθ2 ) and wψ = cot(ψ). Then the forward and
θ2 = sin−1 (−sα sβ ) inverse kinematics can be obtained as shown in Eq. 5 and 6.

where tα = tan(α), cx = cos(x) and sx = sin(x) for x = θ1 , 

θ2 ,α or β . −cψ ± c2ψ + s2ψ s2φ1 − s2ψ s2φ2
θ1 = 2tan−1 ( )
Fig. 5 shows the model of the driving systems. The frames sψ sφ1 + sψ sφ2
a, a , b, b , 0 and 2 have co-located origins at the intersection  (5)
of the axes of rotation of two cylinders as well as the normal cψ ± c2ψ + s2ψ c2φ2 − s2ψ c2φ1
θ2 = 2tan−1 ( )
of the inclined planes of cylinders. In the initial position, the sψ cφ1 + sψ cφ2

575
 
sθ1 cθ2 c12 wψ ± s4θ2 − s2θ2 c212 w2ψ + s2θ2 s2θ1 c2θ2
φ1 = sin−1 ( )
s2θ2 + s2θ1 c2θ2

sθ1 c12 wψ ± c4θ1 s4θ2 − c2θ1 s2θ2 c212 w2ψ + c2θ1 s2θ2 s2θ1
φ2 = sin−1 ( )
s2θ1 + c2θ1 s2θ2
(6)
Fig. 7: The SSSD mode of the joint. During the first second the SSOD
C. Differential kinematics mode is used for choosing a deflecting angle. And the next four seconds
From the derivative process we can obtain the Jacobian the joint is in the SSSD mode. The speeds of the two drives during the two
modes are 15rpm, -15rpm and 15rpm, 15rpm respectively.
matrix. And the Jacobian matrix is used in order to find the
singularity of the joint which is given in Eq. 7 where θ1 and
θ2 are the function of φ1 and φ2 . A singularity exists when θ1 counterbalance the payloads due to gravity. And the average
and θ2 are zero for the determinant of the Jacobian matrix is reduction ratios are 3.6:1 and 2:1 respectively. It is obvious
zero. In this position the joint only permits an instantaneous that the output ability of the joint including payload ability,
yaw motion and prevents an instantaneous pitch motion. But power density has been improved by at least two times.
the singularity can be easy avoided by the unique control
strategy which is described in the following section. III. RESULTS
A. Control strategy
 ∂θ  ⎡ sψ cφ1 cθ1 sψ cφ2
⎤
1 ∂ θ1 − sψ sφ s +c We proposed a new control strategy based on its coupling
J= ∂ φ1 ∂ φ2
= ⎣ sψ sφ1ssψθ1s+c ψ cθ1 1 θ1
sψ sφ2 cθ2
ψ cθ1 ⎦
(7) structure by combining the two motion modes rather than
∂ θ2 ∂ θ2 φ1
∂ φ1 ∂ φ2 sψ cφ2 sθ2 −cψ cθ2 cψ cθ2 −sψ sφ2 sθ2 decoupling the mechanism structure. By this way the pro-
D. Motion modes cess for solving the forward and inverse kinematics can be
simplified into an easier way. And the computing time in
By doing the further observation about the forward kine- MATLAB is reduced to a fifth in comparison to the non-
matics. We obtain two special motion modes: SSDD and simplified method.
SSSD modes. Fig. 6 shows that the joint can deflect in a In the spherical coordinate, considering of arbitrary input
plane under the SSDD mode in which the orienting angle φ1 and φ2 of two drives for the forward kinematics. Then
stays the same but the deflecting angle changes. Fig. 7 shows the equivalent inputs can be solved. We suppose that the
that the joint can rotate in a circle under the SSSD mode in speed of the two modes are n1 , n1 and n2 , −n2 . Hence, the
which the deflecting angle α stays the same but the orienting simplification relation can be expressed as
angle β changes with the same speed as the drives. Finally
we can use these two modes to achieve the complex space
trajectory. n1 + n2 = φ1
(8)
n1 − n2 = φ2
So we get

n1 = (φ1 + φ2 )/2
(9)
n1 = (φ1 − φ2 )/2
In the SSSD mode,we obtain

π
β = n1 −
2 (10)
Fig. 6: The SSDD mode of the joint. The speed of the two drives are 15rpm α = α0
and -15rpm respectively.
where α0 is the initial value of orienting angle.
In comparison to other type of single and multi-DOF In the SSOD mode, we get
joints, this joint can make most of its drives performance.
Because the two drives always work together to generate β = β0
the motion of the joint in these two modes. Referring to the c2n2 c2ψ + s2ψ c2n2 + s2n2 c2ψ (11)
angle velocity of the deflecting angle which can be obtain by α = cos−1 ( )
s2n2 + c2ψ c2n2
taking the derivative of its angle, we infer that the reduction
ratio produced by the joint increases as the joint has a higher where β0 is the deflecting angle after SSSD mode.
deflecting angle. The result is that the joint can provide By substituting Eq. 10 into Eq. 11 and 12, we simplify
more force as it deflecting away from the initial position to the forward kinematics as

576
 φ1 + φ2 π
β= −
2 2
c2(φ1 −φ2 )/2 c2ψ + s2ψ c(φ1 −φ2 ) + s2(φ1 −φ2 )/2 c2ψ (12)
α = cos−1 ( )
s2(φ −φ )/2 + c2ψ c2(φ −φ )/2
1 2 1 2

For the inverse kinematics, the output α and β are known.

Hence, if we only take α into consideration, the input
parameters can be expressed as
cψ (1 − cα )
φ1  = −φ2  = sin−1 ( ) (13)
sψ sα
If we only consider β , we can get
π
φ1  = φ2  = β + (14)
2
Then the simplified inverse kinematics can be shown as

cψ (1 − cα ) π Fig. 9: The experimental setup.

φ1 = −φ2  = β + sin−1 ( )+
sψ sα 2
(15)
c ψ (1 − cα ) π
φ2 = −φ2  = β − sin−1 ( )+ obtain the speed characteristics with the specific PID under
sψ sα 2 the two motion modes. Fig. 10 and 11 show the tracking
Finally we can use the simplified results for the real-time quality of the speed loop feedback and the demand speed are
control shown in Fig. 8. We can use this method to avoid ranging from 1000 rpm to 4000 rpm. Both of speed curves
the singularity by choosing the appropriate solution (φ1 , −φ1 ) are observed in five periods (the detailed description of the
based on the previous positions of these variables. motions in one period are shown in Section III-D). Note that
v1 and v2 are motors speed multiplied by the reducer ratio,
N=66.
Although there has some validation about the speed in
these two figures, the range is small and stable. And this
feature does not change because of the two different motion
modes and speed. It has proven that the joint can work under
the unique control strategy with better performance.

Fig. 8: The solving process of the simplified kinematics, (a) is the simplified
forward kinematics and (b) is the simplified inverse kinematics. Fig. 10: The speed characteristics of the two motors in the SSDD mode.

B. Experiments IV. CONCLUSION

In the real-time control speed characteristics has a great In this paper, a robotic joint with two-DOF coupling
influence in performance for the unique control strategy. motion was presented. In comparison to other types of joints
Fig. 9 shows the experimental setup including the prototype, the proposed joint can perform better on several performance
Maxon epos2 controllers, laptop and power supply. We indexes including payload ability, power density and motion

577
 [12] A. Maity, S. K. Mandal, S. Mazumder, and S. Ghosh. ”Serpentine
robot: An overview of current status & prospect,” in 14th National
Conference on Machines and Mechanisms (NaCoMM-09), Durgapur
(2009): 272-278.
[13] H. Choset, and E. Shammas, ”Orientation preserving angular swivel
joint (Grant),” U.S. Patent 6,871,563, March 29, 2005.
[14] H. Ikeda, and N. Takanashi, ”Joint assembly movable like a human
arm (Grant),” U.S. Patent 4,683,406, July 28, 1987.
[15] N. Takanashi, ”A gait control for the hypre-redundant robot O-RO-
CHI,” in proc. ROBOMEC (1996): 79-80.
[16] H. R. Choi, and S. Ryew, ”Anthropomorphic joint mechanism with
two degrees of freedom,” in Proc. IEEE Int. Conf. on Robotics and
Automation (2000): 1525-1530.
[17] T. Kang, H. Choi, and M. Kim, ”Development of anthropomorphic
robot hand SKK robot hand I,” KSME. J 17.2 (2003): 230-238.
[18] E. Shammas, A. Wolf, and H. Choset, ”Three degrees-of-freedom joint
for spatial hyper-redundant robots,” Mech. Mach. Theory 41.2 (2006):
170-190.
[19] O. Salomon, and A. Wolf, ”Inclined Links Hyper-Redundant Elephant
Fig. 11: The speed characteristics of the two motors in the SSSD mode. Trunk-Like Robot,” J Mech Robot 4.4 (2012): 045001.
[20] H. Asama, M. Sato, L. Bogoni, H. Kaetsu, A. Mitsumoto, and I.
Endo, ”Development of an omni-directional mobile robot with 3 DOF
decoupling drive mechanism,” in Proc. IEEE Int. Conf. on Robotics
ability. The joints workspace is 100◦ in deflecting and ±360◦ and Automation (1995): 1925-1930.
in orienting. We demonstrated the workspace by calculating [21] V. Glazunov, ”Design of decoupled parallel manipulators by means of
the theory of screws,” Mech. Mach. Theory 45.2 (2010): 239-250.
the forward kinematics of the joint. The inverse kinematics [22] C. Yang, Q. Huang, and J. Han, ”Decoupling control for spatial
was presented for the further control of the joint. We also six-degree-of-freedom electro-hydraulic parallel robot,” Rob. Comput.
analyzed the Jacobian matrix of the joint to identify its Integr. Manuf. 28.1 (2012): 14-23.
singular configuration. The counterbalance feature of the
joint was proved for the high payload ability. Moreover, a
new control strategy based on the two basic motion modes
was proposed. The simulation and experiments results had
efficient validated its feasibility.
As a future work, this robotic joint will be used to
develop a new type of manipulator. And the comprehensive
performance of the joint will be proven through the practical
application.

R EFERENCES
[1] M. Feng, Y. Fu, B. Pan, and C. Liu, ”Development of a medical robot
system for minimally invasive surgery,” Int. J. Med. Rob. Comput.
Assisted Surg. 8.1 (2012): 85-96.
[2] R. C. Luo, and C. C. Lai, ”Enriched indoor map construction based
on multisensor fusion approach for intelligent service robot,” IEEE
Trans. Ind. Electron. 59.8 (2012): 3135-3145.
[3] T. Hellstrm, ”On the moral responsibility of military robots,” Ethics
Inf Technol 15.2 (2013): 99-107.
[4] KUKA Roboter GmbH. (2016) KUKA Robot Group. [Online]. Avail-
able: http://www.kuka-robotics.com/en/products/industrial robots/
[5] ABB Asea Brown Boveri Ltd. (2016) ABB Group [Online]. Available:
http://new.abb.com/products/robotics/
[6] T. Morita, F. Kuribara, Y. Shiozawa, and S. Sugano, ”A novel mech-
anism design for gravity compensation in three dimensional space,”
in Proc. IEEE/ASME Int. Conf. Advanced Intelligent Mechatronics
(2003): 163-168.
[7] K. Koser, ”A cam mechanism for gravity-balancing,” Mech. Res.
Commun. 36.4 (2009): 523-530
[8] C. Cho, W. Lee, and S. Kang, ”Static balancing of a manipulator with
hemispherical work space,” in Proc. IEEE/ASME Int. Conf. Advanced
Intelligent Mechatronics. (2010): 1269-1274.
[9] A. Wolf, H. B. Brown, R. Casciola, A. Costa, M. Schwerin, E. Shamas,
and H. Choset, ”A mobile hyper redundant mechanism for search and
rescue tasks,” in Proc. IEEE/RSJ Int. Conf. Intelligent Robots and
Systems (2003): 2889-2895.
[10] H. B. Brown, M. Schwerin, E. Shammas, and H. Choset, ”Design and
control of a second-generation hyper-redundant mechanism,” in Proc.
IEEE/RSJ Int. Conf. Intelligent Robots and Systems (2007): 2603-
2608.
[11] C. Bradley, ”Robotic Arm Calibration and Control 6-DOF Powerball
LWA 4P,” 2014.

578