IEEE Robotics & Automation Magazine - March 2023 - 53

METHOD
The proposed algorithm is analyzed under quasi-static conditions.
The speed of bandaging and moving of the human limb
is slow, and the effect of acceleration is ignored. In addition,
the stretching of the bandage is ignored.
The control block diagram is shown in Figure 3. The control
process includes five parts: 1) the trajectory planning and
posture adjustment strategy, 2) a compliant control algorithm,
3) bandage tension control of the end effector, 4) estimation of
the tension direction, and 5) the following strategy.
TRAJECTORY PLANNING AND POSTURE
ADJUSTMENT STRATEGY
The bandaging trajectory planning is based on the method
presented in our previous work [1]. The algorithm first generates
a basic spiral trajectory according to initial position of
the injured limb. During bandaging, the basic trajectory will
be continuously adjusted according to the movement of the
injured limb. More details of the trajectory adjustment method
are described in the " Following Strategy " section. Figure
4 illustrates that the trajectory of the robot arm end
effector will no longer be a simple spiral trajectory when spiral
bandaging is applied to a swaying arm. The trajectory is
adjusted according to the arm swaying. Figure 4 shows a
comparison of the trajectory of the bandaging when the arm
is swaying and when it is not swaying.
To enlarge the workspace, a strategy of posture adjustment for
the end effector is proposed [Figure 5(b)]. The larger the workspace
of the robot's end effector, the larger the allowed following
range. Unlike the traditional definition of workspace, here, it
refers to the range that the end effector can reach in space while
satisfying the constraint of remaining parallel to the injured arm.
As the first six joints of the robot arm cannot achieve a conMathematical
Model of the SEA 96 mm
θs
θm
θl
fs
(Magnetic Encoder)
Shell
Cover
Exploded View of the Elbow Joint
Angle Sensor
a
192 mm
Spring
Cover
Spring
362 mm
Positioning Magnet
Angle Sensor
Input Flange
Section View of the End Effector
Cover
Output Shaft 1
Input Flange
Output Flange
Input Flange
Positioning Magnet
FIGURE 2. The hardware of the robot arm.
MARCH 2023 IEEE ROBOTICS & AUTOMATION MAGAZINE
53
Input Gear
190 mm
Driving Motor
Positioning
Magnet
Angle Sensor
(Magnetic Encoder)
Output Flange
Spring
Output Shaft 2
(Magnetic Encoder)
tinuous 360° rotation, the workspace defined here will be very
restricted when the posture of the end effector is fully given.
To solve this problem, using the fact that the end effector only
needs to remain parallel to the injured arm and allowing the end
effector to adjust itself along the axial direction can effectively
expand the workspace. The problem then becomes one of using
6 DoF to achieve the constraint of 5 DoF of the end effector, with
the robot arm DoF being equivalent to redundancy. The expansion
of the solution space naturally enlarges the workspace. The
end-effector adjustment rule is described as follows:
i -
ze =
Z
[
\
]
]
]]
]
]
]
r-- peh
- +-
-2
2
2
2
r
r
r+-
peh
az y
az y
az y
az y
tan
tan
tan
2^ pe ,
2^ pe
2
pe
tan2^ pe,
xy z
,,
pe pe pe
^^
^
,
,
pe
pe
h
h
^
^
yz
yz
yz
yz
pe
pe
pe
pe
&
&
&
&
#$
$$
$
#
pe
pe
pe
pe
00
00
00
00
1
1
h
h
h
h
(3)
where^h is the position of the end effector relative
to the injured arm, and zei -
effector according to the strategy.
is the rotation angle of the end
TABLE 2. The compliant joint configuration
characteristics.
JOINT
First
through
fifth
Sixth
NUMBER
OF SPRINGS
3
3
Seventh 1
K
(N/MM)
59
20.4
20.4
A
(MM)
STIFFNESS
(NM/°)
19.05 1.121
16
18
0.273
0.115
ACCURACY
(NM)
0.099
0.024
0.01
Section View of
the Elbow Joint
Driving Motor
IEEE Robotics & Automation Magazine - March 2023

Table of Contents for the Digital Edition of IEEE Robotics & Automation Magazine - March 2023
