IEEE Robotics & Automation Magazine - December 2017 - 22

Figure 3. A Nao robot in a red uniform with its hands behind its back to avoid
collision with an opponent player. (Photo courtesy of RoboCup Federation.)

Figure 4. A game scene of the humanoid AdultSize category. (Photo courtesy of
RoboCup Federation.)

The Sim league includes two categories, two-dimensional (2-D) simulation
on the plane and three-dimensional
(3-D) simulation based on Nao robots
as agents in the 3-D space. In the 2-D
simulation, 13 of 16 participating teams
adopted the game software HELIOS
developed by Dr. Hidehisa Akiyama
from Fukuoka University, Japan. After a
solid game effort, HELIOS2017, Dr.
Akiyama's own team, was victorious in
the 2-D simulation category. In the 3-D
category, the champion team was University of Texas (UT) Austin Villa, led
by Prof. Peter Stone, a speaker at the
artificial intelligence seminar during
RoboCup 2017.
The current SPL uses the humanoid
Nao as a standard platform to compete
purely with software to control Nao robots. Participating teams are separated
22

*

IEEE ROBOTICS & AUTOMATION MAGAZINE

*

into two groups: high-ranking teams
and low-ranking teams. The differences
between the two groups are evident;
high-ranking teams find the ball quickly
and exhibit more stable movements
during the game, while low-ranking
teams find the ball slowly and exhibit
unstable movements. Last year's champion team, B-Human from Germany,
was again the champion this year,
thanks to their superior performance in
every movement. One impressive scene
during the high-ranking teams' games
was when the Nao robots put their
hands together behind their backs to
avoid any collisions with opponent
players and goal poles by shrinking
their shoulder width (Figure 3).
The Humanoid league consists of
three categories depending on the
height: KidSize (40-90 cm), TeenSize

DECEMBER 2017

(80-140 cm), and AdultSize (130-
180 cm). Only this league uses a long
artificial turf field (3-cm height), so, due
to the turf height, many teams had
issues with their robots falling down.
Among them, a German team, NimbRo
(TeenSize) and NimbRo-OP2 (AdultSize) won both championships because of
their lengthy experience in the humanoid league (see Figure 4 for a AdultSize
game). In the KidSize category, the
French team Rhoban Football Club
won the championship because of their
stable movements. The team was also
able to distinguish between two types of
kicks: the toe kick and the inside kick.
The inside kick allowed the robot to
pass the ball more accurately, ultimately
helping the Rhoban Football Club to
secure the win.
Rescue League
The Rescue league consists of two categories: rescue real robot (Rescue-Rob)
and rescue simulation (Rescue-Sim). In
Rescue-Rob, an actual-size disaster
situation is set up, and teams compete
on a number of evaluation items, i.e.,
maneuvering, mobility, dexterity, and
exploration. Robot control is based on
teleoperation or autonomy through the
information acquired by various sensors
attached to the robots. Iranian teams
showed good performance and won
both first and third place.
In Rescue-Sim, based on real city
data (e.g., Nagoya, Japan, has been
added), rescue activities like firefighting, evacuation, and restoration are
evaluated. So far, Iranian teams have
shown good performance, but this
year a Turkish team, YILDIZ, won
the championship.
RoboCup@home
In addition to the existing open platform where any team can participate
with their own robots (open platform
league, with 15 teams), the RoboCup@
home league introduced new standard
platforms: the Toyota Human Support
Robot (HSR) as a domestic standard
platform (DSPL) (seven teams) and
SoftBank Robotics' Pepper as a social
standard platform (SSPL) (ten teams),
working toward new challenges for
Table of Contents for the Digital Edition of IEEE Robotics & Automation Magazine - December 2017
