satyatejachukka/adaptive-traffic-signal

How satyatejachukka/adaptive-traffic-signal works

SeepageRL is a traffic-signal control research application: an interactive simulator and a headless experiment engine use seepage-aware state, constrained multi-profile vector Q-learning, and counterfactual explanations to select and audit signal actions. The benchmark runner evaluates controllers across scenarios and seeds; publication plots summarize benchmark results and seepage physics. Edges follow excerpts or README statements; unsampled simulation wiring is not inferred beyond those documented capabilities.

Main entry points: simulation.py Interactive Simulator · headless_runner.py Headless Environment · rl_state.py Seepage State · run_experiments.py Ablation Runner

Components 13

Traffic Physics

Seepage State rl_state.py
Control Config config.py

Control and Audit

Vector Q Learner rl_agent.py
Vector Rewards reward.py
Counterfactual Audit counterfactual_explainer.py

Simulation Runtimes

Interactive Simulator simulation.py
Traffic and Telemetry UI simulation.py
Headless Environment headless_runner.py
Vector Agent Trainer headless_runner.py

Research Outputs

Ablation Runner run_experiments.py
Benchmark Results experiments/benchmark_results.csv
Research Figures plot_paper_figures.py

Other components

Operator

Connections 18

Operator → Interactive Simulator: runs readme.md
Interactive Simulator → Traffic and Telemetry UI: renders readme.md
Interactive Simulator → Vector Q Learner: selects actions simulation.py
Interactive Simulator → Seepage State: encodes state simulation.py
Interactive Simulator → Counterfactual Audit: shows explanations readme.md
Interactive Simulator → Control Config: uses settings simulation.py
Headless Environment → Control Config: uses settings headless_runner.py
Headless Environment → Seepage State: builds state headless_runner.py
Headless Environment → Vector Q Learner: uses learner headless_runner.py
Vector Agent Trainer → Headless Environment: trains in readme.md
Ablation Runner → Headless Environment: runs scenarios run_experiments.py
Ablation Runner → Vector Q Learner: evaluates policies run_experiments.py
Ablation Runner → Seepage State: uses parameters run_experiments.py
Ablation Runner → Benchmark Results: writes results readme.md
Research Figures → Seepage State: plots physics plot_paper_figures.py
Research Figures → Benchmark Results: reads results plot_paper_figures.py
Vector Rewards → Vector Q Learner: delegates reward reward.py
Vector Rewards → Seepage State: uses parameters reward.py

Drawn by GitDiagram on November 11, 2026 from the repository's file tree, README and source files. Read as Markdown ·