Traffic Light Signalling Control Optimization

This is my take on optimizing traffic light controls as a discrete type problem with genetic algorithms.

How to run

1. Building with Docker

   This will build the project with Docker. This must be done before running the program.

   docker build -t tls_optimization .

2. Running with Docker (Discontinued)

   This will run the project with Docker.

   docker run --rm -v $(pwd):/app -w /app tls_optimization

3. Run GA with PyGad

   This will run the PyGad Implementation of the Genetic Algorithm.

   docker run --rm -v $(pwd):/app -w /app tls_optimization python -m src.algorithms.simple_genetic_algorithm

4. View map statistics

   Will display statistics of the downloaded map and traffic network

   docker run --rm -v $(pwd):/app -w /app tls_optimization python -m src.sumo_setup.statistics

5. Generate network data

   This will generate network data that will have phase durations assigned for individual TLS. This is required step to run the optimization algorithm.

   docker run --rm -v $(pwd):/app -w /app tls_optimization python -m src.sumo_setup.generation

6. Generate map

   This will generate a new map following osm.netccfg configurations.

   docker run --rm -v $(pwd):/app -w /app/src/sumo_setup tls_optimization netconvert -c osm.netccfg

7. Configure Simulator variables

   Configure SUMO Simulator variables when running every simulation. We are now using precalculated routes. this step is not necessary.

   docker run --rm -v $(pwd):/app -w /app/src/sumo_setup tls_optimization bash -c 'python $SUMO_HOME/tools/randomTrips.py -n osm.net.xml.gz -o [name of your routes file].rou.xml'

   You can change from random trips to a specific configuration such as setting the specific number of cars generated per second or setting the total number of cars within every simulation.

8. Discover TLS linkage

   This will discover linkage of TLS by Direct Linkage Empirical Discovery.

   docker run --rm -v $(pwd):/app -w /app tls_optimization python -m src.decomposition.dled_optimizer

9. Execute DG2 Grouping

   This will execute Differential Grouping method as an alternative to linkage discovery. Theoretically it is faster than Embpirical Linkage Learning.

    docker run --rm -v $(pwd):/app -w /app tls_optimization python -m src.decomposition.DG2_grouping

10. Execute IRRG

    docker run --rm -v $(pwd):/app -w /app tls_optimization python -m src.decomposition.IRRG

11. Execute Random Search

    Performs random search of n solutions and m evaluations for each solutions.

    docker run --rm -v $(pwd):/app -w /app tls_optimization python -m src.algorithms.random_search

12. Execute Custom Optimizer

    Runs the Custom Optimizer algorithm which was implemented based on LT-GOMEA.

    docker run --rm -v $(pwd):/app -w /app tls_optimization python -m src.algorithms.custom_optimizer

13. Generate Distance Matrices

    Calculates the network distance matrices (Shortest, Euclidian, Fastest) used by the optimizer.

    docker run --rm -v $(pwd):/app -w /app tls_optimization python -m src.plot.tls_distances_shortest
    docker run --rm -v $(pwd):/app -w /app tls_optimization python -m src.plot.tls_distances_euclidian
    docker run --rm -v $(pwd):/app -w /app tls_optimization python -m src.plot.tls_distances_fastest

14. Generate Clustering Dendrograms

    Generates hierarchical clustering dendrogram plots to visualize the linkage tree structure.

    docker run --rm -v $(pwd):/app -w /app tls_optimization python -m src.plot.plot_dendrograms

15. Analyze Linkage Statistics

    Analyzes the calculated distance matrices to determine the optimal clustering thresholds, calculating max valid non-singleton clusters and the median cluster size for each distance metric.

    docker run --rm -v $(pwd):/app -w /app tls_optimization python -m src.algorithms.linkage_analyzer

16. Run Baseline Configuration

    Evaluates the current configuration of the instance map.

    docker run --rm -v $(pwd):/app -w /app tls_optimization python -m src.algorithms.evaluate_baseline

17. Run LT-GA with PyGad

    This will run the PyGad Implementation of the custom LT - Genetic Algorithm.

    docker run --rm -v $(pwd):/app -w /app tls_optimization python -m src.algorithms.lt_genetic_algorithm

18. Run Differential Evolution (SHADE)

    Runs the EvoX SHADE implementation on the baseline TLS configuration. Behavior depends on NOVEL_MUTATION in config.py: when False, a single plain SHADE run is performed; when True, three runs are executed (one per Ward distance tree: shortest / euclidian / fastest) with end-of-generation pair-cluster mutation applied to a MUTATION_RATE fraction of the population. Requires step 13 (distance matrices) when NOVEL_MUTATION=True.

    docker run --rm -v $(pwd):/app -w /app tls_optimization python -m src.algorithms.differential_evolution

Quick Start Pipeline (Rental Machine Setup)

If you would like to setup the whole pipeline, please run step 1, 5, 6, 13, and 12 in order.

Docker cleaning commands

Docker has build-in commands that are ment to be used for house keeping tasks:

• docker image prune: delete all dangling images (as in without an assigned tag)
• docker image prune -a: delete all images not used by any container
• docker system prune: delete stopped containers, unused networks and dangling image + dangling build cache
• docker system prune -a: delete stopped containers, unused networks, images not used by any container + all build cache