Skip to main content
Formal Specification

Mauritanian desert dodgeball and tag game — Formal Specification

Mauritania · chase-tag · Confidence: medium

← Back to game page

Implementation Suitability

Suitability: engine-readyFidelity: discretizedGap-fills: 2
# Soufla — AI-optimized specification ## General idea A multi-agent corridor evasion and dodgeball game. Team A (Runners) attempts to sprint across a 2D desert corridor between two safe zones while Team B (Fielders stationed along flanking sidelines) passes a ball and hurls raycast projectiles to eliminate crossing runners before they reach the destination base. ## Formal model - **Abstraction taken:** The 2D sandy corridor (20×10m) is modeled with continuous positions and synchronous discrete ticks (Δt=100ms). Speed dynamics, ball flight physics, and dive evasions are evaluated via raycast collision checks and stamina mechanics. - **Player/agent count:** 2 teams (TA=Runners, TB=Fielders), each with K=5 agents. - **Turn model:** Synchronous real-time ticks across multi-inning rounds. ## Formal rules R1. **Corridor Geometry & Initialization:** - Corridor bounded by X∈[0,20], Y∈[0,10]m. - Safe Base 1: X∈[0,3], Y∈[0,10]. Safe Base 2: X∈[17,20], Y∈[0,10]. - Flanking sidelines (Fielders): Y≤1.0m (South flank) and Y≥9.0m (North flank). - All TA agents start in Base 1; TB agents split along North and South flanks. - Ball starts in possession of Fielder 0. R2. **Runner Movement & Evasion:** - TA agents may choose when to leave Base 1 and sprint toward Base 2: - Max sprint speed Vsprint=4.5m/s. - Agent may execute `dive_evade()`: lowers profile (Z→0), halts forward momentum for 0.5s, reducing collision cross-section by 75%. - Reaching X≥17.0m marks a successful safe crossing (+1 point). R3. **Fielder Passing & Throwing:** - Fielder holding ball cannot run (V=0) and may execute: - `pass_ball(target_fielder_id)`: Passes ball to teammate along opposite or same flank. - `throw_at_runner(target_runner_id, power)`: Hurls ball at designated runner. - Ball trajectory is a 2D raycast vector. If ball intersects the bounding radius of a runner in the open corridor: - **Hit Confirmed!** The runner is eliminated (`status = eliminated`). - [gap-fill] If runner executes `catch_ball()` and catches cleanly (Pcatch=f(reflex)), thrower is eliminated and 1 teammate is revived. R4. **Inning Resolution & Role Inversion:** - When all K runners of TA have either crossed safely or been eliminated: - Inning ends. Total safe crossings added to TA match score. - Teams swap roles (TA↔TB), and new inning begins. R5. **Victory Condition:** - The first team to accumulate 10 total crossing points across completed innings wins the match. ## Edge cases - **Tie at the end condition:** If both teams have equal points after a set number of full innings, a 1-runner sudden-death sprint determines the winner. - **A player with no legal move:** Movement, passing, or throwing is always legal for active agents. - **Repetition / cycles:** Bounded by an inning tick limit Tmax⁡=600 ticks (60 seconds). [gap-fill] Any runners remaining in Base 1 or corridor at Tmax⁡ are auto-eliminated. - **Simultaneous claims on the same resource or position:** Fielders stay on flanks; ball possession transfers cleanly via discrete event handling. - **Player elimination mid-round; turn order repair:** Eliminated runners are moved to the bench until the next inning or a catch revival. - **Resource exhaustion:** Exactly 1 ball and fixed 2K agent pool. - **Termination guarantee:** Guaranteed by finite inning time limits and monotonic score progression toward 10 points. ## State representation - **Persistent state:** Match scores (SA,SB), current running team ID, completed innings count. - **Dynamic state:** Inning tick t, ball state (position, velocity, carrier ID), active corridor agent list. - **Agent state:** Position (Xi,Yi), velocity, stamina, role, status ∈{in_base1,sprinting,in_base2,eliminated}. - **Observable information:** Fully observable 2D state for all agents. ## Action space - For Runners: `sprint(dir_x, dir_y)`, `dive_evade()`, `catch_ball()`. - For Fielders: `pass_ball(fielder_id)`, `throw_at_runner(runner_id, power)`, `lateral_shift(dy)`. ## Playability notes - **Single-player-vs-AI:** Highly playable: human controls sprinting runner (timing crossing windows, reading passing lanes, diving under throws) or fielder (orchestrating crossfire passes to trap AI runners). - **Group games (K=5):** Multi-agent cooperative-competitive benchmark. - **Physical-skill games:** Ball aiming and dive timing modeled via 2D physics and reaction margins. - **What is lost in the adaptation:** The physical sensation of soft desert sand flying during diving catches and shouts of cheering spectators. ## Implementation hints - Implement 2D raycast circle-intersection for ball-runner collision detection. - First milestone: Implement 1 Runner sprinting across a 20×10 corridor vs 2 stationary Fielders passing and throwing.

Download as JSON

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

Download JSON
Cite this page

Mauritanian desert dodgeball and tag game (Mauritania). (n.d.). Games of Countries. https://gamesofcountries.com/games/mr-mauritania/soufla

@misc{gamesofcountries-mr-mauritania-soufla,
  title = {Mauritanian desert dodgeball and tag game (Mauritania)},
  year = {n.d.},
  publisher = {Games of Countries},
  url = {https://gamesofcountries.com/games/mr-mauritania/soufla}
}