Definition of Vehicles, Vehicle Types, and Routes

Filename extension .rou.xml
Type of content Vehicles, Vehicle Types, and Routes
Open format? Yes
SUMO specific? Yes
XML Schema routes_file.xsd

There are various applications
that can be used to define vehicular demand for SUMO. Of course it is
also possible to define the demand file manually or to edit generated
files with a text editor. Before starting, it is important to know that
a vehicle in SUMO consists of three parts:

  • a vehicle type which describes the vehicle’s physical properties,
  • a route the vehicle shall take,
  • and the vehicle itself.

Both routes and vehicle types can be shared by several vehicles. It is
not mandatory to define a vehicle type. If not given, a default type is
used. The driver of a vehicle does not have to be modelled explicitly.
For the simulation of persons which walk around or ride in vehicles, additional definitions are necessary.

Initially, we will define a vehicle with a route owned by him only:

    <vType id="type1" accel="0.8" decel="4.5" sigma="0.5" length="5" maxSpeed="70"/>

    <vehicle id="0" type="type1" depart="0" color="1,0,0">
      <route edges="beg middle end rend"/>


By giving such a route definition to SUMO (or
SUMO-GUI), SUMO will build a
red (color=1,0,0) vehicle of type “type1” named “0” which starts at time
0. The vehicle will drive along the streets “beg”, “middle”, “end”, and
as soon as it has approached the edge “rend” it will be removed from the

This vehicle has its own internal route which is not shared with other
vehicles. It is also possible to define two vehicles using the same
route. In this case the route must be “externalized” – defined before
being referenced by the vehicles. Also, the route must be named by
giving it an id. The vehicles using the route refer it using the
“route”-attribute. The complete change looks like this:

    <vType id="type1" accel="0.8" decel="4.5" sigma="0.5" length="5" maxSpeed="70"/>

    <route id="route0" color="1,1,0" edges="beg middle end rend"/>

    <vehicle id="0" type="type1" route="route0" depart="0" color="1,0,0"/>
    <vehicle id="1" type="type1" route="route0" depart="0" color="0,1,0"/>


A vehicle may be defined using the following attributes:

Attribute Name Value Type Description
id id (string) The name of the vehicle
type id The id of the vehicle type to use for this vehicle.
route id The id of the route the vehicle shall drive along
color color This vehicle’s color
depart float (s) or one of triggered, containerTriggered The time step at which the vehicle shall enter the network; see #depart. Alternatively the vehicle departs once a person enters or a container is loaded
departLane int/string (≥0, “random”, “free”, “allowed”, “best”, “first”) The lane on which the vehicle shall be inserted; see #departLane. default: “first”
departPos float(m)/string (“random”, “free”, “random_free”, “base”, “last”, “stop”) The position at which the vehicle shall enter the net; see #departPos. default: “base”
departSpeed float(m/s)/string (≥0, “random”, “max”, “desired”, “speedLimit”) The speed with which the vehicle shall enter the network; see #departSpeed. default: 0
arrivalLane int/string (≥0,”current”) The lane at which the vehicle shall leave the network; see #arrivalLane. default: “current”
arrivalPos float(m)/string (≥0(1), “random”, “max”) The position at which the vehicle shall leave the network; see #arrivalPos. default: “max”
arrivalSpeed float(m/s)/string (≥0,”current”) The speed with which the vehicle shall leave the network; see #arrivalSpeed. default: “current”
line string A string specifying the id of a public transport line which can be used when specifying person rides
personNumber int (in [0,personCapacity]) The number of occupied seats when the vehicle is inserted. default: 0
containerNumber int (in [0,containerCapacity]) The number of occupied container places when the vehicle is inserted. default: 0
reroute bool Whether the vehicle should be equipped with a rerouting device (setting this to false does not take precedence over other assignment options)
via id list List of intermediate edges that shall be passed on rerouting

Note: when via is not set, any <stop>-elements that belong to this route will automatically be used as intermediate edges. Otherwise via takes precedence.

departPosLat float(m)/string (“random”, “free”, “random_free”, “left”, “right”, “center”) The lateral position on the departure lane at which the vehicle shall enter the net; see Simulation/SublaneModel. default: “center”
arrivalPosLat float(m)/string (“left”, “right”, “center”) The lateral position on the arrival lane at which the vehicle shall arrive; see Simulation/SublaneModel. by default the vehicle does not care about lateral arrival position

Repeated vehicles (Flows)

It is possible to define repeated vehicle emissions (“flow”s), which
have the same parameters as the vehicle except for the departure time.
The id of the created vehicles is “flowId.runningNumber” and they are
distributed either equally or randomly in the given interval. The
following additional parameters are known:

Attribute Name Value Type Description
begin float(s) first vehicle departure time
end float(s) end of departure interval (if undefined, defaults to 24 hours)
vehsPerHour float(#/h) number of vehicles per hour, equally spaced (not together with period or probability)
period float(s) insert equally spaced vehicles at that period (not together with vehsPerHour or probability)
probability float([0,1]) probability for emitting a vehicle each second (not together with vehsPerHour or period), see also Simulation/Randomness
number int(#) total number of vehicles, equally spaced
<flow id="type1" color="1,1,0"  begin="0" end= "7200" period="900" type="BUS">
    <route edges="beg middle end rend"/>
    <stop busStop="station1" duration="30"/>


One may notice, that the route itself also got a
color definition, so the attributes of a route

Attribute Name Value Type Description
id id (string) The name of the route
edges id list The edges the vehicle shall drive along, given as their ids, separated using spaces
color color This route’s color
repeat int The number of times that the edges of this route shall be repeated (default 0)
period time (s) When defining a repeating route with stops and those stops use the until attribute, the times will be shifted forward by ‘period’ on each repeat

There are a few important things to consider when building your own

  • Routes have to be connected. At the moment the simulation raises an
    error if the next edge of the current route is not a successor of
    the current edge or if the vehicle is not allowed to drive on any of
    the lanes. If you want the old behavior where a vehicle simply
    stopped at the end of the current edge and was possibly “teleported”
    to the next edge after a waiting time, use the Option –ignore-route-errors.
  • Routes have to contain at least one edge.
  • The route file has to be sorted by starting times. In fact this is
    only relevant, when you define a lot of routes or have large gaps
    between departure times. The simulation parameter –route-steps, which defaults
    to 200, defines the size of the time interval with which the
    simulation loads its routes. That means by default at startup, only
    routes with departure times <200 are loaded, if all the vehicles
    have departed, the routes up to departure time 400 are loaded etc.
    pp. This works only if the route file is sorted. This behavior may
    be disabled by specifying –route-steps 0. It is possible to load
    unsorted route files as an additional file which will load the whole
    file at once.

The first two conditions can be checked using <SUMO_HOME>/tools/route/, the third can be “fixed”
using <SUMO_HOME>/tools/route/


sumo may enter an infinite loop when given an unsorted route file with person definitions.

Incomplete Routes (trips and flows)

Demand information for the simulation may also take the form of origin
and destination edges instead of a complete list of edges. In this case
the simulation performs fastest-path routing based on the traffic
conditions found in the network at the time of departure/flow begin.
Optionally, a list of intermediate edges can be specified with the via
attribute. The input format is exactly the same as that for the
DUAROUTER application and can be found here.

  <trip id="t" depart="0" from="beg" to="end"/>
  <flow id="f" begin="0" end="100" number="23" from="beg" to="end"/>
  <flow id="f2" begin="0" end="100" number="23" from="beg" to="end" via="e1 e23 e7"/>

For more details on how to handle routing errors and influence the
routing in this case see

For supported attributes for flows and trips see here.

Traffic assignement zones (TAZ)

It is also possible to let vehicles depart and arrive at traffic assignment zones (TAZ).
This allows the departure and arrival edges to be selected from a
predefined list of edges. Those edges are used which minimize the travel
time from origin TAZ to destination TAZ. When loading trips into
DUAROUTER the loaded travel times are used (with
empty-network travel times as default). When loading trips into
SUMO, the current travel times in the network are
used as determined by the rerouting

  <trip id="t" depart="0" fromTaz="taz1" toTaz="taz2"/>
  <taz id="<TAZ_ID>" edges="<EDGE_ID> <EDGE_ID> ..."/>


When used in DUAROUTER or SUMO, edge weights within TAZ are ignored.

When loading <taz> in SUMO-GUI the optional attribute shape
can be used to draw an arbitrary polygon border for visualizing the
traffic assignment zone.


When using TAZ with SUMO and DUAROUTER, their edges will be selected to minimize travel time. This is different from TAZ usage in OD2TRIPS where edges are selected according to a probability distribution.

Routing between Junctions

Trips and flows may use the attributes fromJunction, toJunction, and viaJunctions to describe origin, destination and intermediate locations. This is a special form of TAZ-routing and it must be enabled by either setting the SUMO option –junction-taz or by loading TAZ-definitions that use the respective junction IDs. When using option –junction-taz, all edges outgoing from a junction may be used at the origin and all edges incoming to a junction may be used to reach the intermediate and final junctions.

A Vehicle’s depart and arrival parameter

Using the depart... and
arrival...-attributes, it is possible to
control how a vehicle is inserted into the network and how it leaves it.


Determines the time at which the vehicle enters the network (for <flow> the
value of begin is used instead). If there is not enough space in the
network, the actual depart time may be later.

  • When using option –max-depart-delay <TIME> the vehicle is discarded if unable to depart
    after the given delay
  • A random offset to the specified depart time is added for each
    vehicle when using option –random-depart-offset <TIME>
  • When using the special value triggered, the vehicle will depart as
    soon as a person enters it.


Determines on which lane the vehicle is tried to be inserted;

  • ≥0: the index of the lane, starting with
  • random“: a random lane is chosen;
    please note that a vehicle insertion is not retried if it could not
    be inserted
  • free“: the most free (least occupied)
    lane is chosen
  • allowed“: the “free” lane (see above)
    of those lane of the depart edge which allow vehicles of the class
    the vehicle belongs to
  • best“: the “free” lane of those who
    allow the vehicle the longest ride without the need to lane change
  • first“: the rightmost lane the vehicle
    may use

BTW, I like “best” at most – dkrajzew


Determines the position on the chosen departure lane at which the
vehicle is tried to be inserted;

  • ≥0: the position on the lane, starting at the lane’s begin; must be
    smaller than the starting lane’s length
  • "random": a random position is chosen; it is not retried to insert the
    vehicle if the first try fails
  • "free": a free position (if existing) is used
  • "random_free": at first, ten random positions are tried, if all fail, “free” is applied
  • "base": the vehicle is tried to be inserted at the position which lets its
    back be at the beginning of the lane (vehicle’s front
    position=vehicle length)
  • "last": the vehicle is inserted with the given speed as close as possible
  • "stop": if the vehicle has a stop defined, it will depart at the endPos of the stop. If no stop is defined, the behavior defaults to "base"
    behind the last vehicle on the lane. If the lane is empty it is
    inserted at the end of the lane instead. When departSpeed=”max” is set, vehicle speed will not be adapted.


Determines the speed of the vehicle at insertion, where maxSpeed = MIN(speedLimit * speedFactor, vTypeMaxSpeed);

  • ≥0: The vehicle is tried to be inserted
    using the given speed. If that speed is unsafe, departure is
  • random“: A random speed between 0 and
    maxSpeed is used,
    the speed may be adapted to ensure a safe distance to the leader
  • max“: The maxSpeed is used, the speed may be adapted to ensure a safe distance to the leader vehicle.
  • desired“: The maxSpeed is used. If that speed is unsafe, departure is delayed.
  • speedLimit“: The speed limit of the lane is used. If that speed is unsafe, departure is delayed.


Determines the speed at which the vehicle should end its route;

  • current“: the vehicle will not change
    it’s lane when nearing arrival. It will use whatever lane is more
    convenient to reach its arrival position. (default behavior)
  • ≥0: the vehicle changes lanes to end
    it’s route on the specified lane


Determines the position along the destination edge where the vehicle is
conisdered to have arrived;

  • max“: the vehicle will drive up to the
    end of its final lane. (default behavior)
  • <FLOAT>: the position on the lane, starting at
    the lane’s begin; Negative values count from the end of the lane
  • random“: a random position is chosen at
    departure; If vehicle is rerouted a new random position is selected.


Determines the speed at which the vehicle should end its route;

  • current“: the vehicle will not modify
    it’s speed when nearing arrival. It will drive as fast as (safely)
    possible. (default behavior)
  • ≥0: the vehicle approaches it’s arrival
    position to end with the specified speed

A vehicle is defined using the vType-element as shown below:

    <vType id="type1" accel="2.6" decel="4.5" sigma="0.5" length="5" maxSpeed="70"/>

Having defined this, one can build vehicles of type “type1”. The values
used above are the ones most of the examples use. They resemble a
standard vehicle as used within the Stefan Krauß’ thesis.

    <vType id="type1" accel="2.6" decel="4.5" sigma="0.5" length="5" maxSpeed="70"/>
    <vehicle id="veh1" type="type1" depart="0">
        <route edges="edge1 edge2 edge3"/>

This definition is the initial one which includes both, the definition
of the vehicle’s “purely physical” parameters, such as its length, its
color, or its maximum velocity, and also the used car-following model’s
parameters. Please note that even though the car-following parameters
are describing values such as max. acceleration, or max. deceleration,
they mostly do not correspond to what one would assume. The maximum
acceleration for example is not the car’s maximum acceleration
possibility but rather the maximum acceleration a driver choses – even
if you have a Jaguar, you probably are not trying to go to 100km/h in 5s
when driving through a city.

The default car following model is based on the work of Krauß but other
models can be selected as well. Model selection and parameterization is
done by setting further vType-attribures as shown below. The models and their
parameters are described in the following.

    <vType id="type1" length="5" maxSpeed="70" carFollowModel="Krauss" accel="2.6" decel="4.5" sigma="0.5"/>

Available vType Attributes

These values have the following meanings:

Attribute Name Value Type Default Description
id id (string) The name of the vehicle type
accel float 2.6 The acceleration ability of vehicles of this type (in m/s^2)
decel float 4.5 The deceleration ability of vehicles of this type (in m/s^2)
apparentDecel float ==decel The apparent deceleration of the vehicle as used by the standard model (in m/s^2). The follower uses this value as expected maximal deceleration of the leader.
emergencyDecel float ==decel The maximal physically possible deceleration for the vehicle (in m/s^2).
sigma float 0.5 Car-following model parameter, see below
tau float 1.0 Car-following model parameter, see below
length float 5.0 The vehicle’s netto-length (length) (in m)
minGap float 2.5 Empty space after leader [m]
maxSpeed float 55.55 (200 km/h) for vehicles, 1.39 (5 km/h) for pedestrians The vehicle’s maximum velocity (in m/s)
speedFactor float 1.0 The vehicles expected multiplicator for lane speed limits
speedDev float 0.1 The deviation of the speedFactor; see below for details (some vClasses use a different default)
color RGB-color “1,1,0” (yellow) This vehicle type’s color
vClass class (enum) “passenger” An abstract vehicle class (see below). By default vehicles represent regular passenger cars.
emissionClass emission class (enum) “PC_G_EU4” An emission class (see below). By default a gasoline passenger car conforming to emission standard EURO 4 is used.
guiShape shape (enum) “unknown” a vehicle shape for drawing. By default a standard passenger car body is drawn.
width float 1.8 The vehicle’s width [m] (used only for visualization with the default model, affects sublane model)
height float 1.5 The vehicle’s height [m]
collisionMinGapFactor float depends on carFollowModel (1.0 for most models) The minimum fraction of minGap that must be maintained to the leader vehicle to avoid a collision event
imgFile filename (string) “” Image file for rendering vehicles of this type (should be grayscale to allow functional coloring)
osgFile filename (string) “” Object file for rendering with OpenSceneGraph (any of the file types supported by the available OSG-plugins)
laneChangeModel lane changing model name (string) ‘LC2013’ The model used for changing lanes
carFollowModel car following model name (string) ‘Krauss’ The model used for car following
personCapacity int 4 The number of persons (excluding an autonomous driver) the vehicle can transport.
containerCapacity int 0 The number of containers the vehicle can transport.
boardingDuration float 0.5 The time required by a person to board the vehicle.
loadingDuration float 90.0 The time required to load a container onto the vehicle.
latAlignment string center The preferred lateral alignment when using the sublane-model. One of (left, right, center, compact, nice, arbitrary).
minGapLat float 0.6 The desired minimum lateral gap when using the sublane-model
maxSpeedLat float 1.0 The maximum lateral speed when using the sublane-model
actionStepLength float global default (defaults to the simulation step, configurable via –default.action-step-length) The interval length for which vehicle performs its decision logic (acceleration and lane-changing). The given value is processed to the closest (if possible smaller) positive multiple of the simulation step length.

Besides values which describe the vehicle’s car-following properties,
one can find definitions of the assigned vehicles’ shapes, emissions,
and assignment to abstract vehicle classes. These concepts will be
described in the following. Also, you may find further descriptions of
implemented car-following models in the subsection #Car-Following Models.

Speed Distributions

The desired driving speed usually varies among the vehicle of a fleet.
In SUMO this is modeled by a speed distribution using the attributes
speedFactor or speedDev. as explained below.


Since version 1.0.0 speed distributions are used by default (speedDev=”0.1″). In older version, speed distributions had to be defined for every vehicle type to avoid homogeneous speeds (and consequently invalid driving behavior because vehicles would never catch up with their leader vehicle)

Vehicle class specific defaults

When defining a vehicle type with a vClass, the following default speed-deviation will be used.

  • passenger (default vClass): 0.1
  • pedestrian: 0.1
  • bicycle: 0.1
  • truck, trailer, coach, delivery, taxi: 0.05
  • tram, rail_urban, rail, rail_electric, rail_fast: 0
  • emergency: 0
  • everything else: 0.1

Global Configuration

Instead of configuring speed distributions in a <vType> definition (as
explained below), the SUMO-option –default.speeddev <FLOAT> can be used to set
a global default. Seeting this value to 0 restores pre-1.0.0 behavior.

Defining speed limit violations explicitly

Each vehicle has an individual speed factor which is multiplied with the
speed limit (edge speed) to determine the desired driving speed (default
1.0). A vehicle with speed factor 1.2 drives up to 20% above the speed
limit whereas a vehicle with speed factor 0.8 would always stay below
the speed limit by 20%. By setting attributes speedFactor and
speedDev as show below this individual speed factor for all vehicles
of a type can be set to a fixed value.

<vType id="example" speedFactor="1.2" speedDev="0"

Defining a normal distribution for vehicle speeds

The desired driving speed usually varies among the vehicle of a fleet.
While this could be modeled by defining a new type for each vehicle and
assigning a distinct speed factor for each type (as above) this would be
quite cumbersome. Instead the attribute speedFactor can also be used
to sample a vehicle specific speed factor from a normal distribution.
The parameter can be given as “norm(mean, dev)” or “normc(mean, dev,
min, max)”. Using speedFactor=”normc(1,0.1,0.2,2)” will result in a
speed distribution where 95% of the vehicles drive between 80% and 120%
of the legal speed limit. For flows, every inserted vehicle will draw an
individual chosen speed multiplier as well. The resulting values in this
example are capped at 20% of speedFactor at the low end to prevent
extreme dawdling and at twice the recommended speed. A vehicle keeps its
chosen speed multiplier for the whole simulation and multiplies it with
edge speeds to compute the actual speed for driving on this edge. Thus
vehicles can exceed edge speeds. However, vehicle speeds are still
capped at the vehicle type’s maxSpeed.


In order to use mean values below 0.2 or above 2.0, the 4-parameter version must be used to modify the cut-off parameters as well.

Defining a normal distribution (old style)

An alternative way to specify speed distributions is to use numerical
values for speedFactor and speedDev. In this case
speedFactor defines the expected value and speedDev defines the
deviation. When using this style, capping cannot be controlled and will
always default to 20% and 200%. Thus the above example can also be
defined as speedFactor=”1″ speedDev=”0.1″.


When used for pedestrians, the speedFactor attribute is applied directly to the maximum speed of the vType since speed limits are not applicable to pedestrians


If the specified departSpeed of a vehicle exceeds the speed limit and it’s vType has a speedFactor deviation > 0, the indivial chosen speed multiplier is at least high enough to accommodate the stated depart speed.

Vehicle Length

Due to the work on car following models, we decided to use two values
for vehicle length. The length-attribute
describes the length of the vehicle itself. Additionally, the
minGap-attribute describes the offset to the
leading vehicle when standing in a jam.

This is illustrated in the following image:


Within the simulation, each vehicle needs – when ignoring the safe gap –
But only length of the road should be marked
as being occupied.

Abstract Vehicle Class

A SUMO vehicle may be assigned to an “abstract vehicle class”, defined
by using the attribute vClass. These classes
are used in lane definitions and allow/disallow the usage of lanes for
certain vehicle types. One may think of having a road with three lanes,
where the rightmost may only be used by “taxis” or “buses”. The default
vehicle class is passenger (denoting normal passenger cars).

The following vehicle classes exist:

vClass bitmask bit comment
ignoring – (all bits set to 0) may drive on all lanes regardless of set permissions.
private 0
emergency 1
authority 2
army 3
vip 4
pedestrian 5 lanes which only allow this class are considered to be ‘sidewalks’ in NETCONVERT
passenger 6 This is the default vehicle class and denotes regular passenger traffic
hov 7 High-occupancy vehicle
taxi 8
bus 9 urban line traffic
coach 10 overland transport
delivery 11 Allowed on service roads that are not meant for public traffic
truck 12
trailer 13 truck with trailer
motorcycle 14
moped 15 motorized 2-wheeler which may not drive on motorways
bicycle 16
evehicle 17 future mobility concepts such as electric vehicles which may get special access rights
tram 18
rail_urban 19 heavier than ‘tram’ but distinct from ‘rail’. Encompasses Light Rail and S-Bahn
rail 20 heavy rail
rail_electric 21 heavy rail vehicle that may only drive on electrified tracks
rail_fast 22 High-speed-rail
ship 23 basic class for navigating waterways
custom1 24 reserved for user-defined semantics
custom2 25 reserved for user-defined semantics

These values are a “best guess” of somehow meaningful values, surely
worth to be discussed. Though, in parts, they represent classes found in
imported formats. They are “abstract” in the means that they are just
names only, one could build a .5m long bus.


vClass values are mainly used for determining access restrictions for lanes and edges. Since version 0.21.0 they will also affect the defaults of some other vType parameters. These defaults are documented at Vehicle_Type_Parameter_Defaults.

The following vehicle deprecated classes exist for maintaining backward

deprecated vClass replacement
public_emergency deprecated. use ’emergency’
public_authority deprecated, use ‘authority’
public_army deprecated, use ‘army’
public_transport deprecated, use ‘bus’
transport deprecated, use ‘truck’
lightrail deprecated, use ‘tram’
cityrail deprecated, use ‘rail_urban’
rail_slow deprecated, use ‘rail’

Vehicle Emission Classes

The emission class represents a certain emission class. It is defined
using the emissionClass attribute. Possible
values are given in Models/Emissions and
its subsections.


For a nicer visualization of the traffic, the appearance of a vehicle
type’s vehicles may be changed by assigning them a certain shape using
the guiShape attribute. These shapes are
used when setting the drawing mode for vehicles to simple shapes.
The following shapes are known:

  • “pedestrian”
  • “bicycle”
  • “motorcycle”
  • “passenger”
  • “passenger/sedan”
  • “passenger/hatchback”
  • “passenger/wagon”
  • “passenger/van”
  • “delivery”
  • “truck”
  • “truck/semitrailer”
  • “truck/trailer”
  • “bus”
  • “bus/city”
  • “bus/flexible” (8.25)
  • “bus/overland” (8.25)
  • “rail” (24.5)
  • “rail/light” (16.85)
  • “rail/city” (5.71)
  • “rail/slow” (9.44)
  • “rail/fast” (24.775)
  • “rail/cargo” (13.86)
  • “evehicle”
  • “ship”

Some of these classes are drawn as a sequence of carriages. The length
of a single carriage is indicated in parentheses after the type. For
these types, the length of the vehicleType is used as the overall length
of the train (all carriages combined). For example, a vehicle with shape
rail/cargo and length 70m will have 5
carriages. The number of carriages will always be a whole number and no
carriage will be shorter than the length given in brackets but may be
longer to meet the length requirements of the whole vehicle. When
drawing vehicles with raster images, the image will be repeated for each

In addition, one can determine the width of the vehicle using the
attribute width. When using shapes, one
should consider that different vehicle classes (passenger vehicles or
buses) have different lengths. Passenger vehicles with more than 10m
length look quite odd, buses with 2m length, too.


Not all of these named shapes are implemented.

Further parameters can be used to achieve visualization of individual rail carriages

<vType id="rail">
    <param key="carriageLength" value="20"/>
    <param key="carriageGap" value="1"/>
    <param key="locomotiveLength" value="25"/>   

Car-Following Models

The car-following models currently implemented in SUMO are given in the
following table.

Element Name (deprecated) Attribute Value (when declaring as attribute) Description
carFollowing-Krauss Krauss The Krauß-model with some modifications which is the default model used in SUMO
carFollowing-KraussOrig1 KraussOrig1 The original Krauß-model
carFollowing-PWagner2009 PWagner2009 A model by Peter Wagner, using Todosiev’s action points
carFollowing-BKerner BKerner A model by Boris Kerner

Caution: currently under work

carFollowing-IDM IDM The Intelligent Driver Model by Martin Treiber

Caution: Default parameters result in very conservative lane changing gap acceptance

carFollowing-IDMM IDMM Variant of IDMM

Caution: lacking documentation

carFollowing-KraussPS KraussPS the default Krauss model with consideration of road slope
carFollowing-KraussAB KraussAB the default Krauss model with bounded acceleration (only relevant when using PHEM classes)
carFollowing-SmartSK SmartSK Variant of the default Krauss model

Caution: lacking documentation

carFollowing-Wiedemann Wiedemann Car following model by Wiedemann (2-Parameters)
carFollowing-W99 W99 Car following model by Wiedemann, 10-Parameter version
carFollowing-Daniel1 Daniel1 Car following model by Daniel Krajzewicz

Caution: lacking documentation

carFollowing-ACC ACC Car following model by Milanés V. and Shladover S.E.
carFollowing-CACC CACC Car following model by Milanés V. and Shladover S.E.
carFollowing-Rail Rail Model for various train types

Car-Following Model Parameters

Mostly, each model uses its own set of parameters. The following table
lists which parameter are used by which model(s). Details on car-following models and their parameters can be found here.

Attribute Default Range Description Models
minGap vClass-specific >= 0 Minimum Gap when standing (m) all models
accel vClass-specific >= 0 The acceleration ability of vehicles of this type (in m/s^2) Krauss, SKOrig, PW2009, Kerner, IDM, ACC, CACC
decel vClass-specific >= 0 The deceleration ability of vehicles of this type (in m/s^2) Krauss, SKOrig, PW2009, Kerner, IDM, ACC, CACC
emergencyDecel vClass-specific >= decel The maximum deceleration ability of vehicles of this type in case of emergency (in m/s^2) Krauss, SKOrig, PW2009, Kerner, IDM, ACC, CACC
sigma 0.5 [0,1] The driver imperfection (0 denotes perfect driving Krauss, SKOrig, PW2009, Kerner, IDM, ACC, CACC
tau 1 >= 0 The driver’s desired (minimum) time headway. Exact interpretation varies by model. For the default model Krauss this is based on the net space between leader back and follower front). For limitations, see Car-Following-Models#tau). all Models
k Kerner
phi Kerner
delta 4 acceleration exponent IDM
stepping 0.25 >= 0 the internal step length (in s) when computing follow speed IDM
adaptFactor 1.8 >= 0 the factor for taking into account past level of service IDMM
adaptTime 600 >= 0 the time interval (in s) for relaxing past level of service IDMM
security desire for security Wiedemann
estimation accuracy of situation estimation Wiedemann
speedControlGain The control gain determining the rate of speed deviation (Speed control mode) ACC
gapClosingControlGainSpeed The control gain determining the rate of speed deviation (Gap closing control mode) ACC
gapClosingControlGainSpace The control gain determining the rate of positioning deviation (Gap closing control mode) ACC
gapControlGainSpeed The control gain determining the rate of speed deviation (Gap control mode) ACC
gapControlGainSpace The control gain determining the rate of positioning deviation (Gap control mode) ACC
collisionAvoidanceGainSpeed The control gain determining the rate of speed deviation (Collision avoidance mode) ACC
collisionAvoidanceGainSpace The control gain determining the rate of positioning deviation (Collision avoidance mode) ACC
speedControlGainCACC The control gain determining the rate of speed deviation (Speed control mode) CACC
gapClosingControlGainGap The control gain determining the rate of positioning deviation (Gap closing control mode) CACC
gapClosingControlGainGapDot The control gain determining the rate of the positioning deviation derivative (Gap closing control mode) CACC
gapControlGainGap The control gain determining the rate of positioning deviation (Gap control mode) CACC
gapControlGainGapDot The control gain determining the rate of the positioning deviation derivative (Gap control mode) CACC
collisionAvoidanceGainGap The control gain determining the rate of positioning deviation (Collision avoidance mode) CACC
collisionAvoidanceGainGapDot The control gain determining the rate of the positioning deviation derivative (Collision avoidance mode) CACC
CC1 Spacing Time – s W99
CC2 Following Variation – m W99
CC3 Threshold for Entering “Following” – s W99
CC4 Negative “Following” Threshold – m/s W99
CC5 Positive “Following” Threshold – m/s W99
CC6 Speed Dependency of Oscillation – 10^-4 rad/s W99
CC7 Oscillation Acceleration – m/s^2 W99
CC8 Standstill Acceleration – m/s^2 W99
CC9 Acceleration at 80km/h – m/s^2 W99
trainType string id for pre-defined train type Rail

To select a car following model the following syntax should be used:

<vType id="idmAlternative" length="5" minGap="2" carFollowModel="IDM" tau="1.0" .../>

Default Krauss Model Description

The default model is a modification of the model defined by Stefan Krauß
in Microscopic Modeling of Traffic Flow: Investigation of Collision Free Vehicle Dynamics. The
implemented model follows the same idea as that of Krauß, namely: Let
vehicles drive as fast as possibly while maintaining perfect safety
(always being able to avoid a collision if the leader starts braking
within leader and follower maximum acceleration bounds). The implemented
model as in <SUMO_HOME>/src/microsim/cfmodels/MSCFModel_Krauss.cpp has the following differences:

  • Different deceleration capabilities among the vehicles are handled
    without violating safety (the original model allowed for collisions
    in this case)
  • The formula for safe velocity was adapted to maintain safety when
    using the Euler-position update rule. This was done by
    discretizing some of the continuous terms. The original model was
    defined for the Ballistic-position updated rule and would produce
    collisions when using Euler. See also

Lane-Changing Models

The lane-changing models currently implemented in SUMO are given in the
following table.

Attribute Value Description
LC2013 The default car following model, developed by Jakob Erdmann based on DK2008 (see SUMO’s Lane-Changing Model). This is the default model.
SL2015 Lane-changing model for sublane-simulation (used by default when setting option –lateral-resolution <FLOAT>). This model can only be used with the sublane-extension.

Caution: This model may technically be used without activating sublane-simulation but this usage has not been fully tested and may not work as expected.

DK2008 The original lane-changing model of sumo until version 0.18.0, developed by Daniel Krajzewicz (see Traffic Simulation with SUMO – Simulation of Urban Mobility).

Mostly, each model uses its own set of parameters. The following table
lists which parameter are used by which model(s).

Attribute Description Models
lcStrategic The eagerness for performing strategic lane changing. Higher values result in earlier lane-changing. default: 1.0, range [0-inf[ LC2013, SL2015
lcCooperative The willingness for performing cooperative lane changing. Lower values result in reduced cooperation. default: 1.0, range [0-1] LC2013, SL2015
lcSpeedGain The eagerness for performing lane changing to gain speed. Higher values result in more lane-changing. default: 1.0, range [0-inf[ LC2013, SL2015
lcKeepRight The eagerness for following the obligation to keep right. Higher values result in earlier lane-changing. default: 1.0, range [0-inf[ LC2013, SL2015
lcOvertakeRight The probability for violating rules gainst overtaking on the right default: 0, range [0-1[ LC2013
lcOpposite The eagerness for overtaking through the opposite-direction lane. Higher values result in more lane-changing. default: 1.0, range [0-inf[ LC2013
lcLookaheadLeft Factor for configuring the strategic lookahead distance when a change to the left is necessary (relative to right lookahead). default: 2.0, range ]0-inf[ LC2013, SL2015
lcSpeedGainRight Factor for configuring the threshold asymmetry when changing to the left or to the right for speed gain. By default the decision for changing to the right takes more deliberation. Symmetry is achieved when set to 1.0. default: 0.1, range ]0-inf[ LC2013, SL2015
lcSpeedGainLookahead Lookahead time in seconds for anticipating slow down. default: 0, range ]0-inf[ LC2013, SL2015
lcCooperativeRoundabout Factor that increases willingness to move to the inside lane in a multi-lane roundabout. default: lcCooperative, range ]0-1[ LC2013, SL2015
lcCooperativeSpeed Factor for cooperative speed adjustments. default: lcCooperative, range ]0-1[ LC2013, SL2015
lcSublane The eagerness for using the configured lateral alignment within the lane. Higher values result in increased willingness to sacrifice speed for alignment. default: 1.0, range [0-inf] SL2015
lcPushy Willingness to encroach laterally on other drivers. ”default: 0, range 0 to 1 SL2015
lcPushyGap Minimum lateral gap when encroaching laterally on other drives (alternative way to define lcPushy). ”default: minGapLat, range 0 to minGapLat SL2015
lcAssertive Willingness to accept lower front and rear gaps on the target lane. The required gap is divided by this value. ”default: 1, range: positive reals LC2013,SL2015
lcImpatience dynamic factor for modifying lcAssertive and lcPushy. default: 0 (no effect) range -1 to 1. Impatience acts as a multiplier. At -1 the multiplier is 0.5 and at 1 the multiplier is 1.5. SL2015
lcTimeToImpatience Time to reach maximum impatience (of 1). Impatience grows whenever a lane-change manoeuvre is blocked.. default: infinity (disables impatience growth) SL2015
lcAccelLat maximum lateral acceleration per second. default: 1.0 SL2015
lcTurnAlignmentDistance Distance to an upcoming turn on the vehicles route, below which the alignment should be dynamically adapted to match the turn direction. default: 0.0 (i.e., disabled) SL2015
lcMaxSpeedLatStanding Upper bound on lateral speed when standing. default: maxSpeedLat (i.e., disabled) LC2013, SL2015
lcMaxSpeedLatFactor Upper bound on lateral speed while moving computed as lcMaxSpeedLatStanding + lcMaxSpeedLatFactor * getSpeed(). default: 1.0 LC2013, SL2015
lcLaneDiscipline Reluctance to perform speedGain-changes that would place the vehicle across a lane boundary. default: 0.0 SL2015
lcSigma Lateral positioning-imperfection. default: 0.0 LC2013, SL2015

The parameters are set within the <vType>:

<vType id="myType" lcStrategic="0.5" lcCooperative="0.0"/>

Junction Model Parameters

The behavior at intersections may be configured with the parameters
listed below.


These parameters are not available in version 0.30.0 and older

Attribute Value Type Default Description
jmCrossingGap float >= 0 (m) 10 Minimum distance to pedestrians that are walking towards the conflict point with the ego vehicle. If the pedestrians are further away the vehicle may drive across the pedestrian crossing.
jmIgnoreKeepClearTime float (s) -1 The accumulated waiting time (see Option –waiting-time-memory) after which a vehicle will drive onto an intersection even though this might cause jamming. For negative values, the vehicle will always try to keep the junction clear.
jmDriveAfterRedTime float (s) -1 This value causes vehicles to violate a red light if the duration of the red phase is lower than the given threshold. When set to 0, vehicles will always drive at yellow but will try to brake at red. If this behavior causes a vehicle to drive so fast that stopping is not possible any more it will not attempt to stop. This value also applies to the default pedestrian model.
jmDriveAfterYellowTime float (s) -1 This value causes vehicles to violate a yellow light if the duration of the yellow phase is lower than the given threshold. Vehicles that are too fast to brake always drive at yellow..
jmDriveRedSpeed float (m/s) maxSpeed This value causes vehicles affected by jmDriveAfterRedTime to slow down when violating a red light. The given speed will not be exceeded when entering the intersection.
jmIgnoreFoeProb float 0 This value causes vehicles to ignore foe vehicles that have right-of-way with the given probability. The check is performed anew every simulation step. (range [0,1]).
jmIgnoreFoeSpeed float (m/s) 0 This value is used in conjunction with jmIgnoreFoeProb. Only vehicles with a speed below or equal to the given value may be ignored.
jmSigmaMinor float, scaling factor (like sigma) sigma This value configures driving imperfection (dawdling) while passing a minor link (ahead of the intersection after having comitted to drive and while still on the intersection).
jmTimegapMinor float s 1 This value defines the minimum time gap when passing ahead of a prioritized vehicle.
impatience float or ‘off’ 0.0 Willingess of drivers to impede vehicles with higher priority. See below for semantics.

The parameters are set within the <vType>:

<vType id="ambulance" jmDriveAfterRedTime="300" jmDriveAfterRedSpeed="5.56"/>


The impatience of a driver is value between 0 and 1 that grows whenever
the driver has to stop unintentionally (i.e. due to a jam or waiting at
an intersection). The impatience value is computed as

MAX(0, MIN(1.0, baseImpatience + waitingTime / timeToMaxImpatience))

Where baseImpatience is configured by setting the vType-attribute
impatience and timeToMaxImpatience is set using the option –time-to-impatience (default
300s). Setting this option to 0 disables impatience growth. The value of baseImpatience may be negative to slow the growth of
the dynamically computed impatience. It may also be defined with the
value off to prevent drivers from becoming impatient.

The impatience value is used to represent a drivers willingness to
impede vehicles with higher priority. At a value of 1 or above, the
driver will use any gap that is safe in the sense of
collision-avoidance even if it means that another vehicle has to brake
as hard as it can. At a value of 0, the driver will only perform
maneuvers that do not force other vehicles to slow down. Intermediate
values interpolate smoothly between these extremes.

Default Vehicle Type

If the type attribute of a vehicle is not
defined it defaults to "DEFAULT_VEHTYPE".
By defining a vehicle type with this id (<vType id="DEFAULT_VEHTYPE" ..../>) the default parameters for
vehicles without an explicititly defined type can be changed. The change
of the default vehicle type needs to occur before any reference to the
type was made, so basically before any vehicle or vehicle type was
defined. So it should always be at the top of the very first route file.

Instead of defining routes and vTypes explicitly for a vehicle
SUMO can choose them at runtime from a given
distribution. In order to use this feature just define distributions as

Vehicle Type Distributions

    <vTypeDistribution id="typedist1">
        <vType id="type1" accel="0.8" length="5" maxSpeed="70" probability="0.9"/>
        <vType id="type2" accel="1.8" length="15" maxSpeed="50" probability="0.1"/>


The python tool can be used to generate large distributions that vary multiple vType parameters independently of each other.

Using existing types

    <vType id="type1" accel="0.8" length="5" maxSpeed="70" probability="0.9"/>
    <vType id="type2" accel="1.8" length="15" maxSpeed="50" probability="0.1"/>
    <vTypeDistribution id="typedist1" vTypes="type1 type2"/>

Route Distributions

    <routeDistribution id="routedist1">
        <route id="route0" color="1,1,0" edges="beg middle end rend" probability="0.9"/>
        <route id="route1" color="1,2,0" edges="beg middle end" probability="0.1"/>

A distribution has only an id as (mandatory) attribute and needs a
probability attribute for each of its child elements. The sum of the
probability values needs not to be 1, they are scaled accordingly. Note,
that probability defaults to 1.00 when not specified. At the moment
the id for the children is mandatory, this is likely to change in future

A distribution can be used just as using individual types and routes:

    <vehicle id="0" type="typedist1" route="routedist1" depart="0" color="1,0,0"/>


When using DUAROUTER with input files containing distributions, the output files will contain a fixed route and type for each vehicle and the distributions will be gone. This is to ensure that the each vehicles route will fit its sampled vClass when using the input files with SUMO

Vehicles may be forced to stop for a defined time span or wait for
persons by using the stop element either as part of a route or a vehicle
definition as following:

    <route id="route0" edges="beg middle end rend">
        <stop lane="middle_0" endPos="50" duration="20"/>
    <vehicle id="v0" route="route0" depart="0">
        <stop lane="end_0" endPos="10" until="50"/>

The resulting vehicle will stop twice, once at lane middle_0 because of
the stop defined in its route and the second time because of the stop
defined in the vehicle itself. The first stop will last 20 seconds the
second one until simulation second 50. For a detailed list of attributes
to stops see below. For a description on how to use them to simulate
public transport see Simulation/Public Transport.

Stops can be childs of vehicles, routes, persons or containers.

Attribute Type Range Default Remark
busStop string valid busStop ids if given, containerStop, chargingStation, edge, lane, startPos and endPos are not allowed
containerStop string valid containerStop ids if given, busStop, chargingStation, edge, lane, startPos and endPos are not allowed
chargingStation string valid chargingStation ids if given, busStop, containerStop, edge, lane, startPos and endPos are not allowed
lane string lane id the lane id takes the form <edge_id>_<lane_index>. the edge has to be part of the corresponding route
endPos float(m) -lane.length < x < lane.length (negative values count backwards from the end of the lane) lane.length
startPos float(m) -lane.length < x < lane.length (negative values count backwards from the end of the lane) endPos-0.2m there must be a difference of more than 0.1m between startPos and endPos
friendlyPos bool true,false false whether invalid stop positions should be corrected automatically
duration float(s) ≥0 minimum duration for stopping
until float(s) ≥0 the time step at which the route continues
extension float(s) ≥0 the maximum time by which to extend the stop duration due to boarding persons and when waiting for expected persons / triggered stopping
index int, “end”, “fit” 0≤index≤number of stops in the route “end” where to insert the stop in the vehicle’s list of stops
triggered bool true,false false whether a person may end the stop
expected string list of person IDs list of persons that must board the vehicle before it may continue (only takes effect for triggered stops)
expectedContainers string list of container IDs list of containers that must be loaded onto the vehicle before it may continue (only takes effect for triggered stops)
parking bool true,false value of triggered whether the vehicle stops on the road or beside
actType string arbitrary ‘waiting’ activity displayed for stopped person in GUI and output files (only applies to person simulation)
tripId string arbitrary parameter to be applied to the vehicle to track the trip id within a cyclical public transport route
line string arbitrary new line attribute to be set on the vehicle when reaching this stop (for cyclical public transport route)
speed float positive speed to be kept while driving between startPos and endPos
  • If “duration” and “until” are given, the vehicle will stop for at least “duration” seconds.
  • If “duration” is 0 the vehicle will decelerate to reach velocity 0 and then start to accelerate again.
  • If “until” is given and “duration” is not and the vehicle arrives at the stop at or after the time step defined by “until” it will decelerate to speed 0 and then accelerate again.
  • If persons board the vehicle, the stop is extended by the “boardingDuration” of the vehicle or until the “personCapacity” is reached. (or “loadingDuration” and “containerCapacity” for containers).
  • If until is defined in the context of a repeated vehicle insertion (flow) it will be incremented by the difference of vehicle creation time and “begin” of the flow.
  • If neither “duration” nor “until” are given, “triggered” defaults to true. If “triggered” is set to false explicitly the vehicle will stop forever.
  • if “duration” or “until” are given along with “triggered”, then the vehicle will stop until the given duration/until is reached and a person has boarded
  • If “parking” is set to true. The vehicle stops besides the road without blocking other vehicles.


If triggered is true then parking will also be set to true by default. If you then set parking to false you may create deadlocks which prevent the simulation from terminating


Bus stops must have a length of at least 10

startPos and endPos

  • by default vehicles will try to stop and the given endPos
  • if the vehicle comes to a halt earlier (i.e. due to a jam) then the stop counts as reached if the vehicle front is between startPos and endPos
  • if the vehicle picks up a person or container, it can do so as long as the person is between startPos and endPos
  • if the stop uses attribute ‘speed’, than that speed will be maintained between startPos and endPos

A color is defined as red,green,blue or red,green,blue,alpha either
in a vehicle, route or vType.

<route id="r0" color="0,255,255"/>
<type id="t0" color="0,0,255"/>
<vehicle id="v0" color="255,0,0,0"/>

In the default visualization settings the vehicle color will be used if
define, otherwise the type and finally the route color. These settings can be changed.

By default color components should be given as integers in the range of
(0,255) but other definitions are also supported:

color="0.5, 0.5, 1.0"

The transparency value (alpha) only takes effect when also using the vType
attribute imgFile.

Vehicle devices are used to model and configure different aspects such
as output (device.fcd) or behavior (device.rerouting).

The following device names are supported and can be used for the
placeholder <DEVICENAME> below

Automatic assignment

Some devices are assigned automatically. Every <trip> that is loaded into the
simulation is automatically equipped with a rerouting device to
perform the initial route computation.

Other devices such as fcd are assigned automatically when the option –fcd-output
is set.

Assignment by global options

Devices can be configured globally for all vehicles in the simulation by
setting the option –device.<DEVICENAME>.probability (i.e.) –device.fcd.probability 0.25 This will equip
about a quarter of the vehicles with an fcd device (each vehicle
determines this randomly with 25% probability) To make the assignment
exact the additional option –device.<DEVICENAME>.deterministic can be set Another option is to pass the
list of vehicle ids that shall be equipped using the option –device.<DEVICENAME>.explicit <ID1,ID2,…IDk>.


These options take precedence over automatic assignment by output-option.

Assignment by generic parameters

Another option for assigning devices for vehicle types or individual
vehicles is by using generic parameters. This is done by
defining them for the vehicle or the vehicle type in the following way:

    <vehicle id="v0" route="route0" depart="0">
        <param key="has.<DEVICENAME>.device" value="true"/>

    <vType id="t1">
        <param key="has.<DEVICENAME>.device" value="true"/>

    <vehicle id="v1" route="route0" depart="0" type="t1"/>


The <param> of a vehicle has precedence over the <param> of the vehicle’s type. Both have precedence over the assignment by options.