Skip to main content
Formal Specification

Holding of Breath / Touch-and-Tag Combat — Formal Specification

India · chase-tag · Confidence: low

← Back to game page

Implementation Suitability

Suitability: engine-readyFidelity: discretizedGap-fills: 1
# Kabaddi — AI-optimized specification ## General idea Kabaddi is an ancient Indian full-contact territorial invasion sport combining tag, wrestling, and breath-control endurance. Two teams alternate sending a single "raider" into the opposing half of the court to tag defenders while maintaining an unbroken chant (*cant*) within a 30-second clock, before escaping back across the midline against collective defensive tackles. Tagged defenders and pinned raiders are eliminated, with points scored dynamically reviving teammates under classical *Sanjeevani* rules. ## Formal model - **Abstraction taken:** The 13.0 m × 10.0 m court is discretized into a 2D grid of size 26×20 cells (0.5 m resolution per cell). Continuous breath cant and movement are discretized into a 30-second raid clock consisting of 60 ticks (Δt=0.5 seconds). Physical contact and grappling are resolved via discrete state transitions and a parameterized multi-agent collision/tackle resolver. - **Player/agent count:** 2 teams of 7 active players on court (TA,TB). - **Turn model:** Alternating raid cycles across two 20-minute halves (2×40 raids total). ## Formal rules R1. **Court Topography & Zones:** - Midline y=0; Team A half: y∈[−13,0); Team B half: y∈(0,13]. - Baulk lines at y=±3.75 m; Bonus lines at y=±4.75 m; Lobbies at x<−4.0 m and x>4.0 m. - Lobbies are out-of-bounds until initial physical contact occurs between raider and any defender, after which they become active court area for the remainder of that raid. R2. **Raid Execution & Legality:** - Active offensive team sends exactly 1 Raider into the defending half. - Raider must cross the Baulk line (at least 1 foot past line while other foot is airborne), OR execute a physical tag on a defender, to avoid an illegal raid penalty. - **Bonus Point Rule:** If the defending team has 6 or 7 active players on court, and the raider crosses the Bonus line with one foot while lifting the trailing foot off the ground without being held, Toffense is awarded +1 bonus point (bonus does not revive an eliminated player). - **Raid Clock:** Maximum 30 seconds (60 ticks). Failure to return across the midline within 30 seconds eliminates the raider and awards +1 point to the defense. R3. **Contact, Tagging, & Escape:** - Raider attempts actions `hand_touch(defender_id)` or `toe_touch(defender_id)`. - If a tag succeeds: Tagged defender enters state tagged. - **Safe Return:** If the raider crosses the midline (y=0) with any body part while maintaining possession/cant: - All currently tagged defenders are declared "Out" and sent to the waiting block. - Offensive team scores +1 point per eliminated defender. - Each point scored revives 1 previously eliminated offensive teammate in first-in-first-out (FIFO) order (Sanjeevani rule). R4. **Defensive Tackle & Pin:** - Defenders execute coordinated actions: `ankle_hold()`, `thigh_hold()`, `dash()`, or `chain_block()`. - If defenders establish a tackle hold: A struggle contest initiates for up to remaining raid clock ticks. - If defenders hold the raider in the defending half until the 30-second clock expires or the raider's cant breaks: - Raider is declared "Out" and sent to the waiting block. - Defending team scores +1 point and revives 1 eliminated defender. R5. **Lona (All-Out):** - If all 7 players of a team are eliminated (0 players remaining on court): - Opposing team is awarded a **Lona** (+2 bonus points). - All 7 eliminated players immediately re-enter the court with full reset. R6. **Do-or-Die Raid:** - If a team conducts 2 consecutive empty raids (raids scoring 0 touch points and 0 bonus points): - The third raid is a mandatory **Do-or-Die** raid. - The raider must score at least 1 touch point or bonus point; otherwise, the raider is declared "Out" and 1 point is awarded to the defense. R7. **Match Resolution:** - Match consists of two 20-minute halves. - The team with the higher point total at the end of regulation wins. - [gap-fill] In case of a tie in knockout matches, a 5-raid tiebreaker is conducted (each team selects 5 raiders; no bonus line; Baulk line acts as bonus line; baulk crossing mandatory; no player revival). ## Edge cases - **Tie at the end condition:** Handled by R7 (5-raid tiebreaker rules). - **A player with no legal move:** Active raider can always move or attempt an escape; defenders can always reposition. - **Repetition / cycles:** Empty raids are strictly bounded by R6 (Do-or-Die rule on the 3rd empty raid). - **Simultaneous claims on the same resource or position:** In multi-defender collisions, all defenders establishing contact within the tackle frame contribute to the joint tackle strength vector. - **Player elimination mid-round; turn order repair:** Eliminated players leave the court; active team size adjusts dynamically (k∈[1,7]); revivals restore players to the court at the team's back boundary in strict FIFO elimination order. - **Resource exhaustion (pieces, tokens, space, time):** Total raid count and match clock bound all transitions. - **Termination guarantee: why the game provably ends:** Guaranteed by fixed match time (40 minutes) and the finite 30-second clock per raid. ## State representation - **Persistent state:** - Match scores (SA,SB); Match time remaining ∈[0,2400] seconds. - Active rosters: Court sets CA,CB⊆{1,…,7}; Waiting queues WA,WB (FIFO order). - Consecutive empty raid counters (eA,eB)∈{0,1,2}. - Active offensive team ID ∈{A,B}. - **Per-raid ephemeral state:** - Raid clock ticks remaining traid∈[0,60]. - Raider coordinates (xr,yr); Defender coordinates {(xd,yd)}. - baulk_crossed∈bool; bonus_crossed∈bool; lobby_active∈bool. - Set of tagged defenders in current raid Dtagged⊂Cdefend. - tackle_in_progress∈bool. - **Observable information:** Fully observable court state. ## Action space - `move_raider(dx, dy)`: (Raider). - `attempt_tag(technique, target_defender)`: (Raider; technique ∈{toe_touch,hand_touch,kick,dubki}). - `retreat_to_midline()`: (Raider). - `move_defender(defender_id, dx, dy)`: (Individual defender). - `execute_tackle(defender_id, technique)`: (Defender; technique ∈{ankle_hold,thigh_hold,dash,chain_block}). ## Playability notes - **Single-player-vs-AI:** Highly playable as a real-time tactical sport simulation. Human can control the raider's feints and dives against coordinated AI defensive chains, or manage the 7-man defensive unit. - **Group games (n>4):** The 7v7 team game reduces naturally to a 2-agent Commander structure: Offensive Raider vs Defensive Coordinator. - **Physical-skill games:** Physical wrestling is modeled via a contest function: P(escape)=σ(raider_power−∑d∈tacklerstackle_strengthd). - **What is lost in the adaptation:** The physical sensation of heavy lung capacity strain during continuous vocal chanting (*"Kabaddi-Kabaddi"*) and the sting of mat burns. ## Implementation hints - Implement the defensive line as a linked kinematic chain with spring-damper constraints between adjacent defenders to simulate corner-and-cover defensive cohesion. - Branching factor: Raider has 8 movement directions + 4 tag actions; defensive unit has joint movement vectors. Discretizing into role-based macro actions keeps effective branching ≈10–30. - Obvious representational trap: Activating lobbies before physical contact occurs; an un-contacted raider stepping into the lobby is out-of-bounds. - Suggested first milestone: Implement a 1v3 raid with Baulk Line crossing, single-touch tagging, and midline return detection.

Download as JSON

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

Download JSON
Cite this page

Kabaddi (India). (n.d.). Games of Countries. https://gamesofcountries.com/games/in-india/kabaddi

@misc{gamesofcountries-in-india-kabaddi,
  title = {Kabaddi (India)},
  year = {n.d.},
  publisher = {Games of Countries},
  url = {https://gamesofcountries.com/games/in-india/kabaddi}
}