Information Technology Reference
In-Depth Information
The soccer competitions, besides having different rules concerning robots and field
dimensions differ in autonomy of the robots and robot construction details. While in
the Middle-Size and Standard Platform leagues, robots are autonomous and sensors
are mounted on the robots, in the Small-Size league a single agent may decide and
send commands to every robot of a team. The agent of the Small Size league typically
receives information from a camera positioned above the field to detect the robots and
ball positions. The Middle-Size league will be explained in more detail, as some of
the coordination methodologies developed by FC Portugal have been transferred to
the CAMBADA team that competes in this league. In simulated environments there
are several competitions that are detailed in the following sections.
2.1
Middle-Size League
In the Middle-Size League two teams of at most 5 real autonomous robots play soccer
in a 18x12m field. Robots height is limited to 80cm. Robots horizontal shadow must
fit inside a square of 50cm and the weight is limited at 40kg. Robots are completely
autonomous, although they are allowed to communicate each other, and all sensors
must be mounted on the robots. The environment is color marked i.e. the field is
green, the lines and goals are white and the ball used in each competition is an-
nounced at least one month before the tournament. Robots must be black except for
the markers of each team that must be cyan and magenta.
The mechanical and electrical/electronic solutions found to build the robots play a
very important role in the final efficiency to play soccer. Also the vision subsystem is
critical for the final performance of the robots. Games are very active and interesting
in this league and the top teams exhibit some very interesting coordinated behaviour.
2.2
Simulation 2D League
RoboCup Simulation 2D League is one of the 3 leagues that started the RoboCup
official competitions in 1997. In fact a demonstration of the soccer simulator used in
this competition had already been performed during the pre-RoboCup 2006. The view
of the RoboCup Organizers is to focus this league research at the top-level modules of
the soccer robotics problems: the high-level decision and the coordination of teams of,
possibly heterogeneous, robots. Over the years the 2D simulator has evolved, includ-
ing new features and tuning some others, but the core architecture of the simulator is
the same as the one used in 1997.
In the Simulation 2D league a simulator, called soccerserver [42], creates a 2D vir-
tual soccer field and the virtual players, modelled as circles. The simulator imple-
ments the movement, stamina, kicking and refereeing models of the virtual world.
The models in the simulator are a combination of characteristics taken from real
robots (ex: differential drive steering) and from humans (ex: stamina model).
Teams must build the software agents that control each of the 11 virtual robots and
also a coach agent. The control of the agents is performed by sending commands to
the simulator. The main commands are
dash(dPower,dAngle)
,
turn(tAngle)
,
kick(kPower,kAngle)
,
tackle(tAngle)
and
catch(cAngle)
(used only by the goalie). The simulator implements several
