Custom Map Configuration

September 3, 2024 ยท View on GitHub

Environment Setup

  1. Set up the Python environment for MapTR and LimSim.
  2. Install the nuscenes-devkit Python package and modify the map_api.py file. Replace the existing map_api.py located at miniconda3/envs/[env_name]/lib/python3.8/site-packages/nuscenes/map_expansion/map_api.pywith the map_api.py provided in this directory.
  3. Install SUMO on your system.

Workflow

  1. Select a map area from OpenStreetMap: 
    cd ./TrafficManager/networkFiles/
$SUMO_HOME/tools/osmWebWizard.py
    Once completed, the SUMO main interface will open. After confirming that everything is correct, you can close the interface. A folder named with the format yyyy-mm-dd-hh-mm-ss should now appear in the networkFiles/ directory where you ran the command. Navigate to this folder and extract the osm.net.xml.gz file to obtain the osm.net.xml file. Rename the folder to location name, such as location.

2. Generate traffic flow information: Navigate to the folder and generate the traffic flow data.

cd ./TrafficManager/networkFiles/location 
$SUMO_HOME/tools/randomTrips.py -n osm.net.xml.gz -e 300 -r osm.rou.xml -p 0.5

3. Convert the XML file to the NuScenes map data format and save it as a JSON file:

# under DriveArena root dir
python ./TrafficManager/xml_explain2HD.py  --map_name location --root_dir ./TrafficManager/networkFiles

4. Modify the configuration and run: Update the relevant fields in the config.yaml file:

map:
    name: 'location' #'boston-seaport'

Run the code:

    # under DriveArena root folder
    python ./TrafficManager/sim_manager.py