Skip to main content
Formal Specification

Circle Rattan Ball — Formal Specification

Singapore · ball-team · Confidence: high

← Back to game page

Implementation Suitability

Suitability: engine-readyFidelity: discretizedGap-fills: 1
# Sepak Raga Bulat — AI-optimized specification ## General idea Sepak Raga Bulat is a cooperative multi-agent aerial rally game where a ring of 5 agents executes sequential passes and acrobatic volleys to sustain a continuous aerial rally. Each pass involves selecting a strike technique with varying difficulty, target teammate distance, and accumulated team fatigue, terminating when an execution failure grounds the ball. ## Formal model - **Abstraction taken:** 3D ballistic trajectory and biomechanical foot strikes are modeled as discrete pass decisions between 5 fixed nodes arranged in a circle of radius R=2.5 metres. Each pass selects a target node j≠i and technique T, evaluated via a stochastic resolver. - **Player/agent count:** 5 agents in a cooperative circle (Node indices 0..4). - **Turn model:** Dynamic event-driven turns: the agent currently receiving the ball acts next. ## Formal rules R1. State Initialization: 5 agents positioned at angular coordinates ϕk=2πk5 (k∈0..4). Ball starts at agent 0. Turn count K←0, team score S←0, fatigue factor F←0. R2. Receiving & Strike Action: - The receiving agent i may execute up to 2 local control touches (`trap`) before executing a `pass(target_node, technique)`. - Techniques: - `inside_foot` (*sepak sila*): Difficulty D=0.10, style points +1, fatigue +0.01. - `knee_thigh`: Difficulty D=0.15, style points +1, fatigue +0.01. - `head_strike` (*tanduk*): Difficulty D=0.25, style points +2, fatigue +0.02. - `reverse_heel` (*tumit*): Difficulty D=0.45, style points +4, fatigue +0.04. R3. Pass Execution Resolution: - Target distance d=‖x→i−x→target‖. - Success probability: P(success)=clamp(0.05,0.98,skilli×skilltarget−D−0.05×d−F) - If success: K←K+1, S←S+style_points, F←F+ΔF. Active possession transfers to `target_node`. - If failure: The ball is grounded (*mati*). The rally terminates immediately. R4. Round Progression: Each team plays R=3 rounds. R5. Win Condition: In comparative team mode, the team with the highest cumulative rally score S across 3 rounds wins. [gap-fill] In solitary team evaluation, achieving K≥50 consecutive volleys constitutes a grand master victory. ## Edge cases - Tie at the end condition: If two teams tie on total points across 3 rounds, a single sudden-death rally round is played; the team with the longer unbroken rally wins. - A player with no legal move: Not applicable; an agent in possession always has 4 valid teammate target choices and multiple technique options. - Repetition / cycles: Passes naturally cycle among teammates, but the rally terminates with probability 1 due to monotonically increasing fatigue F. - Simultaneous claims on the same resource or position: Possession is transferred strictly to a single designated recipient node. - Player elimination mid-round: If any player faults, the entire team rally ends simultaneously. - Resource exhaustion: Conserved system (1 ball, 5 players). - Termination guarantee: Monotonic fatigue accumulation ensures limK→∞⁡P(success)=0.05, proving finite rally termination. ## State representation - Persistent state: - `round_scores`: array of size 3 (integers ≥0). - `current_round`: integer 1..3. - Per-rally state: - `current_rally_touches`: integer K≥0. - `current_rally_score`: integer ≥0. - `active_ball_node`: integer ∈{0..4}. - `team_fatigue`: float F≥0. - Observable information: Fully observable to all 5 agents. ## Action space - `pass(target_node, technique)`: - `target_node` ∈{0..4}∖{current_node}. - `technique` ∈{inside_foot,knee_thigh,head_strike,reverse_heel}. ## Playability notes - Single-player-vs-AI: 1 human player controls 1 node, coordinating passes with 4 cooperative AI teammates. - Group games (n>4): Adaptable to 4, 6, or 7 agents by adjusting angular node geometry. - Physical-skill games: Foot coordination and acrobatic flexibility are captured in technique difficulties and fatigue mechanics. - What is lost: Three-dimensional body contortions, auditory crisp snap of rattan contact, and shared physical exertion. ## Implementation hints Discrete Markov decision process across 5 nodes. Excellent benchmark for multi-agent cooperative reinforcement learning (MARL).

Download as JSON

Download the full game corpus entry for this game as structured JSON.

Download JSON
Cite this page

Circle Rattan Ball (Singapore). (n.d.). Games of Countries. https://gamesofcountries.com/games/sg-singapore/sepak-raga

@misc{gamesofcountries-sg-singapore-sepak-raga,
  title = {Circle Rattan Ball (Singapore)},
  year = {n.d.},
  publisher = {Games of Countries},
  url = {https://gamesofcountries.com/games/sg-singapore/sepak-raga}
}