Skip to main content
Formal Specification

Khmer Frog Jumping Race (also known as Koun Kangkeb) — Formal Specification

Cambodia · dexterity-small · Confidence: low

← Back to game page

Implementation Suitability

Suitability: engine-readyFidelity: probabilistic-proxyGap-fills: 2
# Kangkeb Lot — AI-optimized specification ## General idea A turn-ticked racing simulation modeling deep-squat frog-jumping locomotion over a 20-meter course. Agents manage physiological fatigue, explosive leap power, and landing balance under the strict rule requirement that both hands must touch the ground between hops. High-effort leaps accelerate track progress at the risk of losing balance or breaking squat posture, triggering a mandatory freeze penalty. ## Formal model - **Abstraction taken:** Continuous squat-jump biomechanics and balance recovery are abstracted into discrete 1D track advances with an effort-allocation decision model and a stochastic overbalance check parameterized by cumulative fatigue. - **Player/agent count:** 2 to N agents (standard: 2–6 racers). - **Turn model:** Simultaneous tick-based simulation (Δt=1 tick). ## Formal rules R1. Course Setup: 1D linear course of length D=20.0 meters. All agents start at position xi=0.0 with `fatigue = 0` and `penalty_timer = 0`. R2. Tick Action Selection: - If `penalty_timer > 0`: The agent is immobilized; `penalty_timer` decrements by 1; `advance = 0.0`. - If `penalty_timer == 0`: The agent simultaneously commits to one leap mode from: - `SHORT_HOP`: Safe jump. Advance Δx=1.0m; Fatigue ΔF=+1. - `POWER_LEAP`: High-propulsion bound. Advance Δx=2.2m; Fatigue ΔF=+4. - `RECOVER_BALANCE`: Cautious reset hop. Advance Δx=0.4m; Fatigue recovers ΔF=−3 (clamped at minimum 0). R3. [gap-fill] Stumble & Foul Mechanics: - Stumble probability is evaluated as P(stumble)=clamp(0.02+0.05×Fi+power_mod,0.0,0.85) (where `power_mod` = 0.0 for `SHORT_HOP`, +0.20 for `POWER_LEAP`, and -0.05 for `RECOVER_BALANCE`). - If a stumble occurs (standing up or landing off-balance): The current tick advance is cancelled (Δx=0.0), and `penalty_timer` is set to 3 ticks [gap-fill: representing the traditional 3-second freeze penalty]. R4. Kinematic Update: For each agent, xi←xi+Δxi, Fi←max⁡(0,Fi+ΔFi). R5. Win Condition: The first agent whose position reaches or exceeds the finish line xi≥20.0 meters wins the race. ## Edge cases - **Tie at the end condition:** If multiple agents cross x≥20.0 on the exact same tick, the agent with greater excess distance xi−20.0 wins; if identical to within 0.01m, a 5-meter sudden-death sprint heat is run. - **A player with no legal move:** All three leap modes are always legal when not in penalty freeze. - **Repetition / cycles (and the loop-breaking rule):** Impossible — track progress xi is monotonically non-decreasing (Δx≥0). - **Simultaneous claims on the same resource or position:** Not applicable — racers run in parallel independent lanes with no spatial overlap. - **Player elimination mid-round; turn order repair:** Not applicable — all racers remain active until crossing the finish line. - **Resource exhaustion:** Fatigue Fi can accumulate up to high values, driving stumble probability toward its maximum cap (85%) and forcing the agent to execute `RECOVER_BALANCE`. - **Termination guarantee: why the game provably ends:** Since Δx>0 for non-stumble actions and P(stumble)<1.0, position xi strictly increases toward D=20.0 meters within finite expected ticks. ## State representation - `positions`: array of floats [x1,…,xN] in [0.0,20.0]. - `fatigue`: array of integers [F1,…,FN]. - `penalty_timers`: array of integers [p1,…,pN]. - `tick_count`: integer. - Observable information: Fully observable — all racer positions and fatigue states are public. ## Action space - `commit_leap_mode(mode)`: where `mode` ∈{SHORT_HOP,POWER_LEAP,RECOVER_BALANCE}. ## Playability notes - **Single-player-vs-AI:** Natural arcade-style risk-reward pacing race against AI competitors with varied risk tolerances. - **Group games (n>4):** Scales trivially to N agents without modification since lanes are independent. - **What is lost:** The physical leg burning, tactile hand-to-dirt contact, and comical visual spectacle of squatting children are abstracted to fatigue management math. ## Implementation hints - Pure Markov Decision Process (MDP) / stochastic race. Optimal policy balances early `POWER_LEAP` bursts with `RECOVER_BALANCE` pacing to keep stumble probability below acceptable risk thresholds. - Implementable in under 50 lines of code.

Download as JSON

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

Download JSON
Cite this page

Khmer Frog Jumping Race (also known as Koun Kangkeb) (Cambodia). (n.d.). Games of Countries. https://gamesofcountries.com/games/kh-cambodia/kangkeb-lot

@misc{gamesofcountries-kh-cambodia-kangkeb-lot,
  title = {Khmer Frog Jumping Race (also known as Koun Kangkeb) (Cambodia)},
  year = {n.d.},
  publisher = {Games of Countries},
  url = {https://gamesofcountries.com/games/kh-cambodia/kangkeb-lot}
}