• Manhattan

Introduction#

This Tutorial explains how to build a Manhattan Mobility Model in SUMO. In this model, a fixed number of vehicles drive randomly on a Manhattan grid network. All files can also be found in the <SUMO_HOME>/docs/tutorial/manhattan directory.

Creating the network#

Creating Manhattan grid networks is supported by the netgenerate application. The option --grid creates grid networks. The number of grid cells can be set using the option --grid.number. There are various options to configure the size and number of the cells and to change the number of lanes and types of junctions. The options for this tutorial are written in a configuration file. The network is created by calling

netgenerate -c manhattan/data/manhattan.netgcfg

Generating vehicles#

The vehicles in a Manhattan mobility model drive randomly according to specified turning ratios. This type of mobility is supported by the jtrrouter application. This application requires <flow>-definitions as input to define the starting point and starting times of vehicles.

Generating random flows for jtrrouter#

The randomTrips.py tool can be used to generated suitable randomFlows with the following options. If the --trip-attributes option is not recognized correctly, try using double quotes around the option value and escape double quotes inside.

 <SUMO_HOME>/tools/randomTrips.py -n net.net.xml -o flows.xml --begin 0 --end 1 \
       --flows 100 --jtrrouter \
       --trip-attributes 'departPos="random" departSpeed="max"'

The option --flows 100 defines the number of vehicles that shall drive in the network. Usually a <flow> is used to define multiple vehicles but in this case each flow generates just a single vehicle at the start of the simulation (hence the option --end 1). The option --jtrrouter must be set to generate flows without destination. Otherwise the generated vehicles might end their trip too early. The arguments supplied to option --trip-attributes are set to ensure that multiple vehicles may enter the source edge in the first step.

The options are also encoded in the script runner.py.

Calling jtrrouter#

The jtrrouter application is called with the generated random flows. To ensure routes of sufficient length the option --allow-loops must be set. Since no sink edge are defined, the option --accept-all-destinations is set. The default turn ratios for the Manhattan Mobility Model (25% right, 50% straight, 25% left) are set via option --turn-defaults 25,50,25.

All options for this tutorial are written in a configuration file. The vehicles are created by calling

jtrrouter -c manhattan/data/manhattan.jtrrcfg

Remarks on Vehicle number#

The number of vehicles in the first few simulation seconds is limited by available road space for vehicle insertions. If the number of vehicles is large with respect to the network size, it may take a few simulation steps before all vehicles have entered the network.

Making vehicles run forever#

Using JTRRouter, routes of arbitrary length can be generated. However, vehicles will eventually reach the end of their route and exit the simulation. To avoid this, the tool generateContinuousRerouters.py can be used.