Like computer chess, robot soccer can be thought of as a game in which the agents on one team cooperate to control more space more comprehensively than the opposing team. Tactical issues such as winning and holding the ball add a complication to the issue of controlling space. As a research experiment we are investigating a ‘space-time possession game’ in which the only issue is the space controlled by the agents individually and in combination. The resultant game is similar to ‘Go’, but the objectives and rules are different. We have begun to study this using cellular automata, and we report on a series of experiments in which the ‘robots’ move at random, move to maximise their individual space, or move to maximise the whole team’s space. Our results suggest the last is the most successful approach.
[1]
M. Castro,et al.
An Algorithm for Robot Path Planning with Cellular Automata
,
2000,
ACRI.
[2]
Blaine A. Price,et al.
Representing Patterns of autonomous agent dynamics in multi-robot systems
,
2003
.
[3]
Martin Müller,et al.
Computer Go
,
2002,
Artif. Intell..
[4]
Jeffrey Johnson.
Visual communication in swarms of intelligent robot agents
,
2006,
Artificial Life and Robotics.
[5]
Tim D. Barfoot,et al.
Coevolving Communication and Cooperation for Lattice Formation Tasks
,
2003,
GECCO.
[6]
Tim D. Barfoot,et al.
Coevolving Communication and Cooperation for Lattice Formation Tasks
,
2003,
ECAL.