Skip to main content
Formal Specification

Red Light (traditional Norwegian stop-and-go stealth racing game; also known as Grønt lys, rødt lys) — Formal Specification

Norway · chase-tag · Confidence: low

← Back to game page

Implementation Suitability

Suitability: adaptation-requiredFidelity: discretizedGap-fills: 9
# Rødt lys — AI-optimized specification ## General idea Rødt lys ("Red Light") is a stop-and-go stealth racing game. A designated Leader faces away at a finish line during "green light" intervals while runners sprint forward. When the Leader calls "Rødt lys!" and spins around, runners must brake instantaneously into complete immobility; any runner detected in motion is penalized by being reset to the start line. The first runner to touch the finish line wins. ## Formal model - **Abstraction taken:** The linear sprint track is discretized into 1D coordinates x∈[0,Dgoal] with Dgoal=25.0 meters [gap-fill]. Time is divided into alternating phases: Green Light (variable duration Tgreen during which runners accelerate and sprint) and Red Light (an instantaneous freeze check where residual velocity v>0 triggers detection). - **Player/agent count:** 3 to 20 agents (1 Leader, N−1 Runners). Standard model: 5 agents (1 Leader, 4 Runners). - **Turn model:** Phase-based round. Continuous or fine-tick movement during green phases, evaluated against Leader detection checks during red freeze triggers. ## Formal rules R1. **Initial Setup:** All N−1 Runners start at xi=0.0 with velocity vi=0.0. The Leader is stationed at x=Dgoal=25.0 facing away (`ORIENTATION_AWAY`). R2. **Green Phase Call:** The Leader selects a green phase duration Tgreen∼Uniform(1.0,4.0) seconds [gap-fill] and begins reciting the cadence. R3. **Runner Sprinting & Deceleration:** During the green phase, runners choose acceleration ai∈[0,amax] up to vmax=6.0 m/s [gap-fill]. When braking, deceleration is bounded by dmax=4.0 m/s2 [gap-fill]. R4. **Red Light Freeze Trigger:** At the end of Tgreen, the Leader calls `RODT_LYS` and transitions to `ORIENTATION_FACING_RUNNERS`. R5. **Inspection & Penalty:** - The Leader inspects all runners. Any runner i whose velocity |vi|>vthreshold (vthreshold=0.05 m/s [gap-fill]) or who commits a balance wobble (P(wobble)=f(braking_intensity) [gap-fill]) is detected. - Any detected runner is penalized: xi←0.0, vi←0.0 (sent back to start). R6. **Victory Condition:** The first runner i to achieve xi≥Dgoal without committing an undetected violation during a red freeze wins the round immediately and becomes the Leader for the next game. R7. [gap-fill] **Max Green Cycles Cap:** If Kmax=30 green-red cycles elapse without a winner, the runner closest to Dgoal wins the round. ## Edge cases - **Tie at the end condition:** If multiple runners cross Dgoal simultaneously during the same green phase, the runner with the largest coordinate xi wins; if identical, shared victory [gap-fill]. - **A player with no legal move:** Not applicable; runners choose acceleration/braking dynamically. - **Repetition / cycles:** Resets to x=0 can create cycles; bounded strictly by the max cycle cap Kmax=30 [gap-fill]. - **Simultaneous claims on the same resource or position:** Multiple runners can share the same longitudinal coordinate x; lane collisions are ignored in the 1D model. - **Player elimination mid-round; turn order repair:** No elimination; penalized runners reset to x=0 and remain active. - **Resource exhaustion:** No physical resources consumed; track distance is fixed. - **Termination guarantee:** Guaranteed by monotonic approach conditioned on conservative braking, backed by the hard cycle limit Kmax=30. ## State representation - **Persistent state:** `leader_id` ∈[0,N−1], `match_scores` (rounds won per agent). - **Per-round state:** `cycle_count` ∈[1,Kmax], `phase` ∈{GREEN,RED_INSPECTION}, `leader_orientation` ∈{AWAY,FACING}. - **Per-runner state:** `position` xi∈[0,Dgoal], `velocity` vi≥0, `status` ∈{ACTIVE,PENALIZED,FINISHED}. - **Observable information:** Runners observe Leader cadence timing (with stochastic duration noise) and own kinematic state. Leader observes full runner velocities only during the instantaneous `RED_INSPECTION` window. ## Action space - `set_acceleration(a)`: Runner control input a∈[−dmax,amax]. - `initiate_call(duration)`: Leader action selecting green light duration. ## Playability notes - **Single-player-vs-AI:** Highly playable as a timing, risk-management, and stopping-distance calculation challenge against an unpredictable AI Leader. - **Group games (n>4):** Scales effortlessly to 20 runners. 2-agent duel (1 Leader vs 1 Runner) works directly with high tension. - **Physical-skill games:** Neuromuscular braking and balance control are parameterized via maximum deceleration limits and balance fault probabilities. - **What is lost in the adaptation:** Physical somatic holding of breath, playground giggles, and micro-postural balancing. ## Implementation hints - Model runners as 1D kinematic integrators: x˙=v,v˙=a. - Implement as a sequential MDP for the runner where the core trade-off is late braking (higher speed, risk of overshooting into red light) vs. early conservative braking. - Suggested first milestone: Implement the 1D kinematic physics loop and Leader random timer trigger, verifying that over-speeding triggers the x=0 reset.

Download as JSON

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

Download JSON
Cite this page

Red Light (traditional Norwegian stop-and-go stealth racing game; also known as Grønt lys, rødt lys) (Norway). (n.d.). Games of Countries. https://gamesofcountries.com/games/no-norway/rodt-lys

@misc{gamesofcountries-no-norway-rodt-lys,
  title = {Red Light (traditional Norwegian stop-and-go stealth racing game; also known as Grønt lys, rødt lys) (Norway)},
  year = {n.d.},
  publisher = {Games of Countries},
  url = {https://gamesofcountries.com/games/no-norway/rodt-lys}
}