Skip to main content
Formal Specification

Baibalaa (Tag-and-Escape Circle Combat) — Formal Specification

Maldives · chase-tag · Confidence: low

← Back to game page

Implementation Suitability

Suitability: adaptation-requiredFidelity: discretizedGap-fills: 3
# Baibalaa — AI-optimized specification ## General idea Baibalaa is an asymmetric tag-and-escape arena combat game set inside a circular perimeter. A single raiding agent enters a bounded ring to tag one of multiple defending agents and must escape across the boundary before the defenders can swarm and immobilize them. The strategic depth lies in the raider's feinting and spatial positioning versus the defending team's coordinated encirclement and trapping formations. ## Formal model - **Abstraction taken:** The continuous physical sprinting, evasion, and grappling on loose sand is abstracted into discrete spatial-tactical rounds comprising approach, tag attempt, and a multi-agent grapple resolution check. - **Player/agent count:** 2 teams; each raid pits 1 raiding agent against k defending agents (where k∈[1,8] active inside the ring). - **Turn model:** Simultaneous discrete ticks within a bounded raid duration (modeled as up to 30 ticks per raid). ## Formal rules R1. Setup: k defenders occupy coordinates within a circular arena of radius R; the raider starts at the perimeter boundary. R2. On each tick, all active agents choose movement vectors within their maximum speed parameters. R3. Legal action set per tick: - Raider: `move(dx, dy)`, `attempt_tag(target_id)`, `evade()`, `dive_exit()`. - Defenders: `move(dx, dy)`, `intercept(raider_id)`, `assist_grapple(target_id)`. R4. Tag Resolution: When the raider is within distance dtag≤1.0 of a defender and executes `attempt_tag`, a tag check succeeds with probability P(tag)=f(speed_delta,agility). Once tagged, the state transitions to Escape Phase. R5. Grapple Resolution: If tagged, defenders within contact range execute `assist_grapple`. The probability of restraining the raider before crossing the boundary is P(pin)=logistic(∑defender_strength−raider_escape_power). R6. Raid Outcomes: - Raider Success: If the raider crosses radius R without being pinned, the tagged defender is permanently eliminated from the set of active defenders. - Defender Success: If the raider is pinned or fails to escape before tick 30, the raider is eliminated. R7. [gap-fill] Raid rotation: Teams alternate single raids; if an attacking raider fails, the defending team receives 1 point. R8. [gap-fill] Game termination: A game concludes when all defenders of a team are eliminated (win by shutout) or after 10 alternating raids per side, with the higher point total winning. ## Edge cases - **Tie at the end condition:** [gap-fill] If points are tied after 10 raids, a sudden-death single raid is played where the team with fewer active defenders raids against an equal-sized defending unit. - **A player with no legal move:** Not applicable; all agents always have valid movement or stance actions. - **Repetition / cycles:** Bounded by the strict 30-tick per-raid timer. - **Simultaneous claims on the same resource or position:** Defenders cannot occupy the exact same spatial grid point; physical collision displacement is resolved by momentum. - **Player elimination mid-round:** When an agent is eliminated, they are removed from the active roster; the next raid simply initializes with the remaining defender subset. - **Resource exhaustion:** Not applicable; arena boundaries and player rosters are finite. - **Termination guarantee:** Provably terminates because each raid lasts ≤30 ticks and the match is bounded by a fixed maximum of 20 raids. ## State representation - **Persistent state:** Team rosters, player elimination status (boolean per player), match score, current attacking team ID. - **Per-turn ephemeral state:** Current tick counter (1..30), agent coordinates (x,y), current raider velocity, tagged status flag, active grappler list. - **Observable information:** Fully observed spatial state; positions, velocities, and elimination rosters are public to all agents. ## Action space - `move(dx, dy)`: Continuous 2D vector clamped to maximum agent movement speed. - `attempt_tag(target_id)`: Precondition: within tagging radius dtag of `target_id`. - `grapple()`: Precondition: within contact range dcontact of raider during Escape Phase. ## Playability notes - **Single-player-vs-AI:** Highly engaging; player controls the lone raider maneuvering against coordinated AI swarm agents, or controls the anchor defender coordinating group tackles. - **Group games (n>4):** Naturally multi-agent; easily scaled by adjusting defender count k. - **Physical-skill games:** Physical contact and sand footing are mapped into speed, acceleration, and strength parameters. - **What is lost:** Real-time muscular balance, grip breaking, and tactile wrestling nuances are collapsed into stochastic logistic checks. ## Implementation hints Low-to-medium complexity. The primary challenge is pathfinding and convex-hull interception logic for the defending AI swarm. Suggested milestone: build a 2D kinematic simulation with a heuristic defender flocking algorithm before implementing full grappling resolution.

Download as JSON

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

Download JSON
Cite this page

Baibalaa (Tag-and-Escape Circle Combat) (Maldives). (n.d.). Games of Countries. https://gamesofcountries.com/games/mv-maldives/baibalaa

@misc{gamesofcountries-mv-maldives-baibalaa,
  title = {Baibalaa (Tag-and-Escape Circle Combat) (Maldives)},
  year = {n.d.},
  publisher = {Games of Countries},
  url = {https://gamesofcountries.com/games/mv-maldives/baibalaa}
}