Introduction to RoboCup 3D Soccer Simulation League
The 3D Soccer Simulation League, a part of RoboCup, is scheduled to run from June 24th to 27th. This league, inaugurated in 2004, focuses on simulated soccer matches, emphasizing the control of humanoid robots.
League Overview
The 3D Soccer Simulation League is a component of the broader RoboCup Soccer Simulation League, comprising two sub-leagues: the 2D Simulation League and the 3D Simulation League.
2D Simulation League: High-level Strategies
In the 2D Simulation League, research centers on high-level strategies, artificial intelligence, and soccer tactics. Represented as small circles within the simulator, the robots are not physical entities. The league primarily examines robot movement, cooperation, and strategic planning within a multi-agent system.
3D Simulation League: Bridging Reality and Simulation
Contrary to the 2D league, the 3D Simulation League acts as a bridge between the 2D Simulation League and real soccer robot competitions. Utilizing humanoid robot models inspired by the Nao robot, teams engage in 11 vs 11 matches. Unlike real robot leagues, the 3D league employs a larger number of players.
Challenges Faced by Teams
Teams in the 3D Simulation League confront two primary challenges. Firstly, they must deploy high-level AI for multi-agent coordination and strategic planning, akin to the 2D simulation. Additionally, teams are tasked with low-level control of robots, managing each joint and motor responsible for movement. This mirrors the challenges encountered in controlling real humanoid robots.
Investigating AI Developments in Humanoid Robots
A fundamental objective of the league is to explore advancements in AI applicable to humanoid robots.
Base Code and Adaptation
Teams often start with shared base codes, with notable examples being those from Team UT Austin Villa (USA) and Team magmaOffenburg (Germany). These base codes, available in C++ and Java, serve as frameworks for team development. Teams can utilize any programming language of their choice, drawing inspiration from these base codes.
Code Sharing Dynamics
While not obligatory, teams are encouraged to share their code. High-performing teams like UT Austin Villa and magmaOffenburg typically release their base codes, facilitating newcomers’ entry into the league.
Educational Resources
In an effort to support new teams, the league has developed a series of tutorial sessions, providing step-by-step guidance on simulator setup and base code utilization. These tutorials are publicly accessible, aiding teams in their initial setup and development process.
New Team Participation
The 3D Soccer Simulation League saw the inclusion of a new team from South Africa this year, alongside the return of several teams after a hiatus.
Adaptations for Virtual Format
Transitioning to a virtual format for RoboCup this year had minimal impact on the simulation leagues. Matches are live-streamed on YouTube, enabling teams and spectators to engage remotely. While maintaining existing rules and simulators, the absence of physical interaction affects the exchange of knowledge and experiences among teams.
Continuous Development and Rule Evolution
Annual revisions to simulators and rules aim to enhance league development and offer teams more challenging gameplay.
Simulator Modifications for 2024
This year, adjustments were made to how fouls are handled in the game to address simulator limitations. Instead of players being transported off the field for certain fouls, they are now moved to a safe distance to mitigate collisions. A new challenge introduced involves incorporating software from the 2D league to interpret commands like “kick” for precise movements, bridging the gap between 2D and 3D simulation leagues.
Future Direction: Simulator Enhancement
The Humanoid League operates on a customized simulator, prompting interest from the 3D league. Plans to transition to a more robust simulator, potentially aligning with the one used by the Humanoid League, aim to bring the 3D league closer to physical competitions.
Inter-League Collaboration
As an executive member with a longstanding tenure, understanding each league’s role within RoboCup and its alignment with overarching goals is paramount. Emphasis is placed on league interaction and contribution to the global federation.
Participation Numbers
This year, the 3D Simulation League boasts 12 participating teams.
Competition Format
The competition unfolds in several stages:
- Preliminary Round: Teams are divided into two groups of six, engaging in intra-group matches to determine seeding for subsequent rounds.
- Round One: Marks the commencement of official classification matches, with one team eliminated from each group.
- Round Two: Comprising two groups of five teams, culminating in the elimination of the two lowest-ranked teams.
- Round Three: Features two groups of four teams; the top two from each group advance to the semi-finals, followed by a final match.
- Ranking Matches: Teams ousted earlier in the competition partake in additional rounds to determine their final rankings, providing extra gameplay opportunities for development.
Duration of Competition
The primary competition unfolds from June 24th to 27th, with supplementary challenges on June 23rd. Teams have the flexibility to modify their codes throughout the event, often dedicating late-night efforts to enhancements. Submission of updated code to the organizing committee at least 30 minutes before the subsequent round allows for real-time adjustments, fostering significant improvements observed among teams during the competition.