The simulation league was originally created to foster research on soccer game playing at a higher level than was previously possible in the real world because of the contemporary restraints of robotics technology.
While it started out as a two dimensional simulation, later on the 3D simulation competition (2004) improved on this by adding an extra dimension. Thereby increasing the realism of the simulated environment and making it more comparable to other (real world) RoboCup league environments; but still being able to simulate more players on the field than would be practical in, say, the middle size league or even the small size league. Thus giving rise to higher level research on larger groups of slightly more realistic soccer playing robots. But it were still only sphere shaped agents running around the field.
Finally as of 2006 the league simulates 3D models of humanoids, based on robots used in the non-simulated RoboCup humanoid league. This opens up opportunities for research on higher level control of humanoid robots as well as higher level behaviors in humanoid soccer and getting still closer to how humans play the game.
In reality, all these changes have caused a slight shift in the aim of the 3D simulation competition over the past few years.
It was initially (and still is in the 2D competition) about finding out how to create higher strategic behaviors in soccer playing multi agent systems, now the changes to the competition has lead it to a slightly lower level of problem solving temporarily. This is because of the increased difficulty of the 3D environment and the problems of controlling humanoid agents at a more basic level.
Notwithstanding these difficulties in the past few years, we are happy to say that the 3D simulation competition - now using simulated humanoid robotic agents- is slowly getting back to a higher level and is yet again ready for research and development of higher level tactical behaviors in the multi agent soccer research area.