IEEE Robotics & Automation Magazine - December 2018 - 87

dual-arm or multiarm space robot can be employed instead of
a single-arm one to extend dexterity and flexibility.
With a dual-arm space robot, one arm can be used as a
mission arm and the other arm as a balanced arm to compensate for the disturbance caused by the motion of the mission
arm on the base [4], [5]. Instead, by making use of the full
potential of robot arms, the two arms can both perform as
mission arms so as to cooperate to implement complex space
tasks. Yoshida et al. discussed two control strategies in [6]:
independent control of each arm and dual-arm coordinated
control for a dual-arm space robot. For the coordinated operation, one arm traces a path, and the other arm works to
maintain the satellite attitude while minimizing joint torques.
The performance of a model-based control algorithm is compared with the transposed Jacobian algorithm in [7] for a
multiarm space robot. The transposed Jacobian algorithm
was able to achieve acceptable control accuracy with a
reduced computational burden.
Adaptive control schemes were proposed [8]-[10] to
achieve coordinated control of the base's attitude and the arm's
motion in joint space. However, for many space missions, the
desired hand trajectory is specified in task space. Because linear parametrization of the space robot model in task space
cannot be realized [11], an adaptive control method cannot
be directly applied. On the other hand, when considering system uncertainties, the joint displacement cannot be accurately
derived from the desired hand position prior to the maneuvers because the mapping from task space to joint space is
dependent on dynamic system parameters that are not accurately known under such a condition. This implies that the
performance of the end effector's motion might be poor even
if the joints can follow their desired trajectories.
Therefore, in this article, we focus on coordinated control
of the base's attitude and the manipulator motion in task
space. Sliding mode control (SMC) has the advantage of coping with system uncertainties and does not require linear
parametrization of the model. But, because of its chattering
effect, a conventional SMC method is not applicable to the
reaction wheels that provide control torques in spacecraft attitude-tracking operations. Thus, a higher-order sliding mode
controller has been developed.
In addition, a main contribution of this article is our application of a novel control methodology, AVSC, to a dual-arm
space robot. This AVSC method can achieve fast set-point
tracking and spacecraft attitude regulation. Through numerical simulation, the developed AVSC controller provides higher control accuracy and reduced energy consumption
compared to higher-order SMC for both path-tracking and
set-point regulation cases. Moreover, AVSC has shown greater
robustness when system uncertainties are included.
Space Robot Model
To represent a typical operation scenario in an on-orbit servicing mission, a space robot is assumed to approach two fixture points of a target with corresponding end effectors.
Detailed assumptions are clarified as follows:

1) The space robot has two arms mounted on its mobile base,
and each arm comprises three rigid links connected by revolute joints, as shown in Figure 2. Then, the robot consists
of seven rigid bodies, with i = 0 representing the spacecraft base and L (i k) (i = 1, 2, 3, k = 1, 2) representing the
ith link of arm k.
2) As a result of assumption 1, the space robot
Because linear
has 6 + 6 degrees of
freedom (DoF), among
parametrization of the
which 6 DoF originate
from the arm motion
space robot model in task
and 6 DoF denote the
spacecraft attitude and
space cannot be realized,
its translation. The six
joint motors generate
adaptive control method
torque to control the
arm motion. For simcannot be directly applied.
plicity, three orthogonally mounted reaction
wheels generate regulation torques along the roll, pitch, and yaw axes in the base
frame, while the translation of the spacecraft base is
not controlled.
3) Because the operations are performed in close proximity
and within a short time compared with the orbital radius
and orbital period, the effects of orbital mechanics are
neglected. In addition, throughout the approaching operation, the target stays within the workspace of the space
robot via the setting of an appropriate initial configuration
for the space robot. As a result, no singular configuration
is directed.
4) The space robot has two identical arms mounted centralsymmetrically with respect to the centroid of the base. Initially, the x B - y B, x G - y G, x T - y T , and x I - y I planes

Phase 1

Phase 2

Observing and Planning

Approaching

Phase 4

Phase 3

Postcapture

Capture

Figure 1. The four phases of space robot operation.

december 2018

IEEE ROBOTICS & AUTOMATION MAGAZINE

IEEE Robotics & Automation Magazine - December 2018

Table of Contents for the Digital Edition of IEEE Robotics & Automation Magazine - December 2018