drtOnline.py#
The drtOnline.py tool allows you to simulate shared demand responsive transport (DRT).
The tool uses TraCI and the taxi device to control
the requests and the drt vehicles. Requests arrive dynamically and are handled by
a dispatcher that seeks to combine multiple requests in order to maximize the number
of requests served while minimizing the mileage of the entire fleet of vehicles.
If you don't need to simulate a shared service, you can directly use the taxi device in sumo and don't need this tool. See Taxi and the respective Taxi service tutorial .
The tool requires python3 and the python LP (Linear Programming) modeler PuLP at least in version 2.0. Please do not install the python3-pulp package on ubuntu 20.04 (or earlier) use pip instead.
The minimal call is:
python tools/drt/drtOnline.py -n <net-file> -r <route-file> --taxi <vehicle-file>
The route file contains the requests, which are defined as persons with ride elements with the lines of the drt vehicles. For example:
<routes>
<person id="0" depart="0.00">
<ride from="edgeStart" to="edgeEnd" lines="taxi"/>
</person>
</routes>
The vehicle file contains the drt vehicles, which must be equipped with a taxi device. The file should looks like:
<routes>
<vType id="taxi" vClass="taxi">
<param key="has.taxi.device" value="true"/>
</vType>
<trip id="taxiA" depart="0.00" type="taxi" personCapacity="12">
<stop lane="edge0" triggered="person"/>
</trip>
</routes>
You can change the capacity as well as other attributes of the vehicles just like in sumo. See vehicle attributes.
As output a tripinfo.xml file is generated with information about each request or person
and vehicle. Persons who have as result the attribute <vehicle="NULL">, represent orders
that could not be served and therefore were rejected.
The tool has different options to adapt the drt service. The most relevant are: the --max-wait option which defines the maximum waiting time for pick-up the requests (default 900 s). With --sim-step the step time to collect new reservations is defined (default 30 seconds). The --drf option defines the direct route factor that is used as an auxiliary to set the maximum travel time with the drt service that a person can accept. The maximum travel time with the drt service cannot be greater than the direct travel time multiplied by this factor, which by default has a value equal to 2.
darpSolvers.py#
This is an auxiliary tool used by the drtOnline.py that contains the algorithms available for the routing of the drt vehicles.
drtOrtools.py#
The tool drtOrtools.py models demand responsive transport (DRT) in SUMO via
TraCI and the taxi device and uses
ortools to solve the vehicle routing problems.
As drtOnline.py, requests arrive dynamically and multiple requests are
combined in order to maximize the number of requests served while minimizing
the mileage (or duration) of the entire vehicle fleet. Due to the use of ortools as
the solver, further extensions are easier to model.
The tool requires python and the packages numpy and ortools. The minimum call is:
python drtOrtools.py -s example.sumocfg
with example.sumocfg being a valid SUMO configuration file which refers to
a network file, routes etc. The routes files should contain taxis with taxi devices
and persons (or containers) with rides including lines="taxi".
Also pre-booking is possible.
drtOrtools.py has the following options:
| Attribute Name | Value Type | Default | Description |
|---|---|---|---|
| -e --end | int (seconds) | end time in .sumocfg else 90000s |
time step to end the simulation |
| -i --interval | int (seconds) | 30s | time interval for collecting requests, if there are new requests after this interval the solver starts |
| -n --nogui | bool | false | if true, uses only the command line version of sumo, else uses sumo-gui |
| -v --verbose | bool | false | prints more information about the dispatch calculation |
| -t --time-limit | int (seconds) | 10s | maximum time for the solver to calculate a single dispatch problem |
| -d --cost-type | str('distance' or 'time') | distance | defines if distance or time is used as costs for the dispatch |
| -f --drf | float | 1.5 | direct route factor: factor for maximum allowed cost for the route of a request in relation to the minimum possible cost (of the direct route); use '-1' to ignore this constraint |
| -a --fix-allocation | bool | false | if true: as soon as a solution was found and a request was assigned to a taxi, this assignment keeps up for all following calculations |
| -w --waiting-time | int (seconds) | 900s | maximum waiting time until service after spontaneous booking or the earliest pick-up time; use '-1' for ignoring the waiting time |
| -p --penalty-factor | 'dynamic' or int | 'dynamic' | factor for different penalties of constraints in relation to the routing costs; 'dynamic' prefers holding constraints of already accepted requests, than accepting new requests |
For further help on the command line arguments run:
python drtOrtools.py --help
The solver requires a matrix with costs to travel
between drop-off locations, pick-up locations and current vehicle positions.
Costs can be distances or travel times depending on the option --cost-type.
The costs are obtained during simulation by calls to traci.simulation.findRoute().
After a regular interval (option --interval), open requests are assigned to
vehicles of the DRT fleet and the cost matrix is passed to the auxiliary file
ortools_pdp.py, where the vehicle routing problem is solved using ortools.
The solution is translated and applied with traci.vehicle.dispatchTaxi.