IEEE Robotics & Automation Magazine - December 2011 - 54

*
Table 2. Technical specifications of the dolphin
robot prototype.
Items

Characteristics

Dimension (L 3 W 3 H)

$560 mm 3 240 mm
3 160 mm

Total mass

$3.29 kg

Mass of the moving slider

$0.23 kg

Drive mode

dc servomotors

Onboard sensors

IR detector and pressure
sensor

Power supply

7.4 V rechargeable
Li-polymer batteries

Working voltage

Operation time

$2h

Control mode

Radio frequency
(433 MHz)

Maximum forward speed

$ 1.1 BL/s (in open
waters)

Minimum turning radius

$ 0.4 BL

Multicamera

Ball
Dolphin Robots
Public
Swim Pool
(Max. Length: 7 m
Wireless
Max. Width: 3.5 m)
Transmitter

Figure 7. A schematic of the exhibition system.

Forward Speed (cm/s)

45
Amplitude I
Amplitude II
Amplitude III

40
35
30
25
20
15
10
10

14 16 18 20 22 24
Angular Frequency (rad/s)

Figure 8. Swimming speed evaluation by modulating the
oscillation frequency and/or amplitude of the joints at its
posterior body, where Amplitude I ¼ fA2 ¼ 6, A3 ¼ 8, A4 ¼ 10,
ci ¼ 2g, Amplitude II ¼ fA2 ¼ 6, A3 ¼ 8, A4 ¼ 10, ci ¼ 3g, and
Amplitude III ¼ fA2 ¼ 6, A3 ¼ 8, A4 ¼ 10, ci ¼ 4g.

IEEE ROBOTICS & AUTOMATION MAGAZINE

DECEMBER 2011

for human-dolphin interaction) and the motion status information (e.g., position, orientation and velocity) of the
dolphin robots. Although color serves as the basic feature for
identifying the environment information by a conventional
dual-camera system, a set of infrared (IR) LEDs (see Figure
6) instead of color patches is used for the dolphin robot.
According to prescribed arrangements of the LEDs such as
the trio and the quintet, every robot is associated with a specific LED identification (ID) via a custom switch. The movements of the robots can then be captured by two cameras,
with the IR filters being placed in front of the lens. Thus, a
nonlinear state estimator composed of online adaptive segmentations and closure operations running on the host PC
produces an estimate of the environment and the robots.
Simultaneously, a combination of the two-dimensional
(2-D) position information and the measured depth of the
robot will further yield a 3-D position for more complex
demonstration tasks. From a command and control perspective, with the multicamera tracking system providing realtime feedback, a closed loop, in either an individual or cooperative manner, is formed. Notice that commands either
from the human operator or from the intelligent control
console will be sent to the robots via a wireless connection.
Laboratory Tests
Before putting the dolphin robots into service, systematic
tests on its swimming performance and control method
have been conducted in a laboratory swim tank. The
dimension of the tank is approximately 1, 680 mm 3
870 mm 3 800 mm (length 3 width 3 height). An overhead vision-measuring system is also applied to provide a
performance estimation of the robots. In the following
prototype test, the used dolphin robots, though in a somewhat different configuration, have the same mechatronic
structure as the final exhibition robot.
Swimming Forwards
Through iterative speed measurements, one can obtain the
influence of the characteristic parameters on the propulsive speed. Without special declaration, each experimental
data point corresponds to an average of five repeated trials.
Figure 8 plots a comparison of time-averaged forward
speeds (v) with different frequencies (x) and amplitudes
(A). Specifically, three groups of amplitudes were examined, while the frequency was varied in each case. It is
observed that v increases directly with x and A for a moderate x, whereas a larger A does not lead to a higher v for a
larger x, e.g., x ¼ 20 rad/s. This phenomenon partly suggests the adopted servomotors hardly track the desired
oscillations within the higher frequencies area while
affording enough torques. On the other hand, the experimental results correlate well with the biological fact: the
heave amplitude appears to increase with speed at low
swimming speeds while it is independent of swimming
speed at routine and sprint speeds, with a constant proportion of BL (approximately 20%) [3].

Table of Contents for the Digital Edition of IEEE Robotics & Automation Magazine - December 2011