IEEE Robotics & Automation Magazine - December 2020 - 23

box and helped move the object outside the box once the item
was grasped. Without assistance from the legs, the arm
released the shell after trying to force its way out of the con-
fined environment.

of the two bending segments are directly related. As a result,
solving the inverse kinematics of the soft manipulator with
opposing curvature requires computing only geometric func-
tions when modeling the whole manipulator. Compared with
the previous opposite-bending-and-stretching structure mod-
eling method [28] (with a computational time of 8.2 ms), the
computational time (11 μs) of the current modeling method
is notably shortened.
Meanwhile, with the addition of the bending of the third
soft segment, the workspace of the current soft manipulator
(500 mm in length, 520 mm in width, and 304 mm in height)
is greatly enhanced compared with the previous version
(260 mm in length, 240 mm in width, and 220 mm in height).
Therefore, this method enhances the practical use of the
manipulator for underwater grasping in the natural environ-
ment. Note that there is a tradeoff between the reduced defor-
mation configuration space and the computational complexity.
This tradeoff depends on the application backgrounds. We
expect that this method will have a broader range of applica-
tions in the future: it may shed light on the modeling of other
soft continuum robots that use actuation approaches other

Discussion
Our study investigated the interaction between a completely
soft arm and a legged robot in a real-world underwater envi-
ronment. In this section, we discuss the experimental results
of the underwater manipulation, including both force and
precision, and the application of those features for more com-
plicated mission-like underwater experiments.
Simplifying the Soft Manipulator's Inverse
Kinematics Problem
Solving higher-order nonlinear equations and training a prac-
tical kinematics model in oceanic environments remain chal-
lenging [28]. In this article, we proposed a simple and inverse
kinematics solution for a soft manipulator-the oppositebending S shape. The structure design offers advantages for
the kinematic modeling of the manipulator since the attitudes

120

Power

Power (W)

90
60
30
0
5

6

Depth

Roll

4

-300
-450

15
Time (s)
(a)

20

Pitch

25

300

2
0

-4
5

10

15
Time (s)
(b)

20

25

200
100

-2

-600

400

Yaw
d (mm)

-150

Angle (°)

Depth (mm)

0

10

5

1) Grasp Object

10

15
Time (s)
(c)

2) Legs Move

20

25

0
(d)

3) Release Object

Object
Target Area
Start Point

Object

Object
(e)

Figure 9. The results of the mobile pick and place. The (a) power consumption, (b) gripper depth, (c) body orientation, and (d) object
placement offset distances, representing all the data collected during the experiments. (e) The yellow, pink, and purple backgrounds
represent the grasping, standing, and releasing phases, respectively.

DECEMBER 2020

*

IEEE ROBOTICS & AUTOMATION MAGAZINE

*

23
IEEE Robotics & Automation Magazine - December 2020

Table of Contents for the Digital Edition of IEEE Robotics & Automation Magazine - December 2020
