Trajectory Modelling for Vessel Tracking and Collision Avoidance

• Author: Xing Yifan

• Vessel-Trajectory-Modeller is a python machine learning package which receives large vessel maneuvering datasets from Automated Identification System (AIS). It makes use of Machine learning techniques for trajectory clustering and it extracts minumum maneuvering distance between pairs of vessels for modelling the interation among the different vessels. Please cite the papers listed at the end if you use this code

• Setup Module Requirements:

  • Python

  • Numpy

    pip install numpy
    

  • Scipy

    pip install scipy
    

  • Scikit Learn

    pip install -U scikit-learn    
    

  • matplotlib

    pip install matplotlib
    

• Usage:

  • run inputDataSearcher.py

    • To search for raw {IMO}.csv files from the dataSource/dynamic folder and create a sub folder structure with the {vessel_type} input
      • dataSource/dynamic is excluded in the repository, contact xingyifan@u.nus.edu for the complete files
    • To specifiy a maximum number of .csv files copied, type a number at the second time asking for input (or press enter for searching for all .csv files of that {vessel_type})
    • a folder structure of {vessel_type}/input/*.csv will be created

  • run readCSV.py

    • input the {vessel_type} from previous step as root_folder asked
    • the raw {IMO}.csv will be cleaned
    • a sub folder under {vessel_type}/cleanData/*.csv will be created

  • run trajectory_modeller.py

    • input the {vessel_type} from previous step as root_folder asked

    • choose 'y' to 'Need to compute min_distance_matrix for vessel interaction? (y/n)' if min distance matrix which contains information on vessel interaction is not yet computed

      • min distance matrix is a {n} by {n} matrix for {n} vessels, where each entry in the matrix is the minimum maneuvering distance that the two vessels will keep from each other

    • trajectory_modeller will continue with endpoint extraction, trajectory formation, trajectory interpolation and trajectory clustering

    • Final subfolder structure will be as the following:

      • {vessel_type}/

        |___ cleanedData/

        - Contains the cleaned {IMO}.csv files after cleaning errorneous points from the raw input
        |___ {IMO}.csv 
        

        |___ endPoints/

        - Contains the learned endpoints for each vessel data file
        |___ {IMO}_endpoints.csv
        

        |___ input/

          - Contains the raw input from AIS
          |___ {IMO}.csv 
        

        |___ trajectories/

          - Contains the trajectory learned for each vessel data file, from one endpoint to another.
          |___ {IMO}_trajectory_endpoint_{endpoint index}_to_{endpoint index}.csv 
        

    • Compiled learning results will be found in ./{vessel_type}LearningResult/ (Includes the following files)

      • protocol_trajectories_with_cluster_size.csv
        • A list of trajectory patterns learned stored with their corresponding width (pattern size)
      • endpoints_to_protocol_trajectories.csv
        • A dictionary of [Endpoint: Possible trajectories that start from this endpoint]
      • mmsi_list.csv
        • The list of MMSI identifier (unique indentifier) for all the vessels involved in the learning procedure
      • vessel_min_distance_matrix.csv
        • The learned minimum maneuvering distance matrix between pairs of vessels
      • vessel_speed_to_distance.csv
        • A dictionary of [A pair of vessels' maneuvering relative speed: A pair of vessels' maneuvering distance]

• Please cite this paper if you use this code

  Pedrielli, G., Xing, Y., Peh, J.H., and and Ng, S.H. A Real Time Simulation Optimization Framework for Vessel Collision avoidance and the case of Singapore Strait. Working paper, 2016.