Improving Emergency Vehicle Routing with Additional Road

Features

Danny Brady

and Byungkyu Brian Park

Data Science Insitute, University of Virginia, Charlottesville, VA 22904, U.S.A.

Department of Civil and Environmental Engineering, University of Virginia, Charlottesville, VA 22904, U.S.A.

Keywords: Emergency Vehicle Routing, Shortest Path, Dijkstra Algorithm, Road Features.

Abstract: Routing decisions affect everybody traveling on the roadways, but perhaps those decisions are never more

important than for emergency responders. In a field where seconds count, getting lost or taking a sub-optimal

route can mean the difference between life and death. Routing methods have advanced from hand drawn maps

to smart phones with dynamic routing using GPS, and new data science methods can improve it further. This

paper employs a modification to a common routing algorithm to improve route decision for emergency

responders. By considering roadway geometry and connected vehicle technologies that simple routing

algorithms ignore, Dijkstra’s algorithm was able to be customized specifically for emergency response. The

lane count, intersection control devices, and median construction were three additional elements of data

incorporated into the modified model. Through a case study of an area of Albemarle County, Virginia, USA

and discussion with domain experts, this paper demonstrates the feasibility and potential benefits of using

such an algorithm to improve routing decisions for emergency responders.

1 INTRODUCTION

Emergency vehicle response technology has not been

changed significantly over the last decade. Ten years

ago, the fire service used paper map books to look up

addresses, and then it was up to the individual

responding to figure out the best route to the

emergency incident. Today it is not much different,

except that an address is typed in to software on a

laptop which shows the location on a map on the

computer. Firefighters are required to use their best

discretion, local area knowledge and a bit of

guesswork to determine how best to get to an

incident. With the current level of mobile computing

power, we propose that its capabilities should be

harnessed to aid in determining the best driving route

to emergency incidents.

There are numerous routing systems available to

the average consumer, from built-in vehicle

navigation to small dash-mounted units and personal

smart phones. While these systems are useful in

aiding transportation, there is additional information

that these systems do not take into account that could

prove extremely valuable to emergency vehicle

operators. These additional features allow the

creation of a routing system specifically tailored to

emergency vehicle response which should aid the in

the speed and safety of their travel. This paper

proposed to utilize additional features on the

transportation network such as the number of lanes,

intersection control devices, and median

construction, and to demonstrate the proposed

shortest path finding algorithm modified from

Dijkstra’s algorithm is suitable for emergency vehicle

route guidance.

2 METHODOLOGY

Standard route guidance systems use a relatively

small number of features for their road network.

Simple systems have information on link length and

speeds, and their connections to other links. Since

road geometry data is not easily accessible or accurate

for freely online, we use tools like Google Maps to

create an accurate representation of the road network

served by Albemarle County’s Seminole Trail Fire

Station, in Virginia, USA. The novelty of this system

is the various other road geometry features and

Connected Vehicle technologies taken into account

than simple systems are capable of analyzing.

Brady, D. and Park, B.

Improving Emergency Vehicle Routing with Additional Road Features.

In Proceedings of the International Conference on Vehicle Technology and Intelligent Transport Systems (VEHITS 2016), pages 187-194

ISBN: 978-989-758-185-4

187

Included are features such as road medians that other

vehicles can use to move out of the way and

intersection pre-emption systems. Details of the road

network used are in the Custom Road Network

section.

The route guidance algorithm was implemented

using a Python program and Javascript was used for

visualizations. Dijkstra’s algorithm, published in

1959, is a standard way to find the shortest distance

on a graph. It utilizes a link weighting that determines

the shortest “distance”, and if that weight is simply

the distance between road intersections the true

shortest distance route can be found. However, we

have created a novel weighting, or cost, by using the

additional road network features so that the route

chosen is not necessarily the shortest distance, but the

most efficient for emergency vehicles. In order to

determine how these additional features should affect

the costs, we have used our first-hand knowledge of

emergency responses as well as other domain experts

in the area.

2.1 Algorithm Implementation

For the route guidance system, Dijkstra’s algorithm

was implemented into a Python code developed in

this paper. The algorithm was originally created to

find the shortest number of nodes between two points

on a graph, but has been modified to consider the

“cost” of each leg. This modification is key, because

it allows the algorithm to optimize the path for

various factors, not only distance. The algorithm

reads in the road network data developed in this paper

as described in the Custom Road Network section,

and uses any cost functions, also implemented in

Python. The algorithm can be seen in the Python

Code section in Appendices, and the use of cost

functions is discussed in more depth in the Cost

Functions section.

2.2 Configuration

A vital step in the creation of a model or simulation is

configuration. Configuration is the process of testing

and adjusting a model to make it to match real world

scenarios. To initially configure the algorithm and

road network, a cost function that only used distances

and road speeds was employed. When used in

conjunction with Dijkstra’s algorithm, this network

yields results that mimic any other standard route

guidance system. We tested several routes and was

able to confirm that the routes generated using this

simple cost function were typical shortest routes.

2.3 Road Network Visualization

Another important aspect of model creation is the use

of visualizations. When creating a custom road

network, a method to check for errors is to create a

map. We tested this using the road network data we

developed in this paper along with a simple webpage

we created. The result is a map of the major roads in

Albemarle County around the 29 North corridor. An

open source javascript library called D3.js was used

to layout our map as shown in Figure 1.

The lengths of each link represent the actual

length between the indicated intersections. The color

of each intersection represents the type of intersection

control device there, either a stop sign (orange), a

traffic light (light blue), or a traffic light with

emergency vehicle pre-emption technology (dark

blue). The thickness of each link also becomes greater

as the speed limit increases. Using this map to help

configure our network was very useful, as it is easy to

see prominent area features. For example, the lines

are thickest on 29, which has a speed limit of 45 mph.

Those intersections also have traffic light pre-

emption, while the smaller streets located in

neighborhoods tend to have stop signs.

Figure 1: Roadway Network in Albemarle County.

2.4 Custom Road Network

There are is no open-source road network information

available that is adequate for this paper. While there

are freely available collections of road network data,

they are currently not well suited for use in route

guidance. It was determined that it would be

satisfactory for the purposes of this paper to create

custom road network data. Maps of the 29 North

corridor were used as a base to determine which roads

to include in the custom network. Smaller roads,

especially dead ends are excluded because no matter

the route taken to get to the smaller road, there is

typically only one route to the final destination from

VEHITS 2016 - International Conference on Vehicle Technology and Intelligent Transport Systems

188

there. This has the added benefit of keeping the

network simpler and needing less custom data.

The road network data was stored using

JavaScript Object Notation (JSON). It contains two

main sections, the intersections and the roads of the

network. Figure 2 shows examples from the actual

network and the format used. The network contains

all of the standard routing information including

names of roads, their connections, their lengths in feet

and their speed limits. Google maps distance

measuring tool and street view were vital in gathering

this information.

ExampleIntersections:

{"name":"29 @ Rio

Rd","controlType":"preemptionSig

nal"}

{"name":"Station

8","controlType":"stop"}



ExampleRoadLink:

{"source":"29@HiltonHeights

Rd","target":"29@PoloGrounds

Rd", "speed":45, "distance":2400,

"lanes":8,"median":"hard"}

Figure 2: Road Network Structure.

Additional features were required in order to improve

route guidance for the specific use case of emergency

responses. These features play a role in emergency

response while not being very important for a typical

driver. Information on the intersection control device,

the number of total lanes on a link, as well as the

median construction were added. The intersection

control device was labeled as either a stop sign, traffic

signal, or a pre-emption signal. Emergency vehicles

could be equipped with Connected Vehicle

technology that signals the infrastructure to change

the light to green in the direction of their approach.

These devices are general only installed on major

roads as they can be costly to install and maintain, and

hold less value on smaller roads with less vehicle

volume.

The number of lanes does not normally play a role

in routing for typical vehicles but is important for

emergency vehicles. When an emergency vehicle is

approaching, other drivers must move out of the way

as quickly and carefully as possible. A road with eight

lanes, such as Route 29, has more area for the vehicles

to move to than a two lane road. It is therefore

preferable for an emergency vehicle to travel on roads

with more lanes if possible. Similarly, soft medians

allow for other drivers to move out of the way even

on a smaller road. This again is preferable to hard or

non-existent medians that would involve a vehicle

driving up over a curb of some kind.

2.5 Cost Functions

The first cost function created calibrated the route

guidance system using the road’s length to calculating

a route’s cost. This results in the true shortest path,

but is more simplified than standard commercial route

guidance algorithms. Using this cost function, we

were able to quickly confirm that the code developed

in this paper was working properly and that

configuration was completed.

The second cost function implemented was one

used in a commercial route guidance system. This

implementation not only uses the road length, but

weights it using the speed limit. This function should

give the path with the shortest travel time under

normal conditions. Some of the newer and more

advanced software, such as Google Maps Directions

and Waze, even account for current traffic conditions.

The last and most important cost function created

was to optimize emergency vehicle response. This

function builds off of the shortest travel time route,

and adds additional factors for analysis. This function

was modified by adding weighting factors based on

each road’s and intersection’s unique features. In

order to determine appropriate weight factors, and to

assess practical implementation, this model was used

in a case study of Seminole Trail Volunteer Fire

Department (STVFD) response area. Domain experts

from STVFD along with some trial and error were

used to create these factors.

3 CASE STUDY

Albemarle County and the city of Charlottesville

provide an interesting area for this case study. There

are various types of roads and intersections with a mix

of features and technologies ranging from small two-

lane rural highways such as Barracks Road to huge

intersections like Rio Rd at Route 29 with a total of

24 lane approaches. This variety provides a unique

road network to test differences between standard and

improved routing algorithms. In order to customize

the proposed approach to Albemarle County, we

spoke with several domain experts from Seminole

Trail Volunteer Fire Department (STVFD) that all

drive emergency vehicles on a regular basis. Their

insight was vital to coming up with the extra road

features considered most important for emergency

responses. These experts described number of lanes,

type of control device, and median construction as the

Improving Emergency Vehicle Routing with Additional Road Features

189

most important factors. The overwhelming consensus

was that the number of lanes is the most important

factor. To take this into account, the cost function

divides the cost of a link by the number of lanes,

indicating that a 4 lane road is twice as preferable to

a 2 lane road when everything else is equal.

In Albemarle County, the fire engines and

ambulances are equipped with a device called an

“Opticom.” These are lights mounted on the front of

each vehicle that flash in a specific pattern or

frequency, and that light is picked up by receivers

mounted on or near several traffic signals on the most

major roadways. These work with actuated signal

controls to turn the proper approach green before the

emergency vehicle arrives, leading to much easier

traversal of the intersection by allowing for other

vehicles to move through the intersection more easily

than if the light was red.

When a route contained an intersection with a

signal equipped with this technology, the cost of the

link is lowered by 40%, a smaller amount than the

lane count. Intersections with standard traffic signals

were also reduced, but only by 20%. These

intersections have green traffic lights some of the

time when an emergency vehicle is approaching

without pre-emption, so they are still preferable to a

stop sign intersection. Since emergency vehicles

should always stop at stop sign intersections until

clear to proceed, these intersections received no cost

reduction.

The least important factor is the median

construction. From experience driving fire trucks, the

average driver tends to attempt to move out of the way

to the right, and not the left, completely ignoring the

median. Still, if there is a soft median or even a two-

way left turn lane in the middle of the road, it is

typically better than a hard median or no median at

all. If other vehicles try to move to the right as far as

they can but there is no shoulder, these soft medians

can be used by the emergency vehicle to get around

more easily. The median construction type received

the smallest cost adjustment in this system. The

domain experts agreed that the least preferable type

of median was a hard median, because even though

most emergency vehicles could drive up over a curb,

it generally is not done and could damage the vehicle.

Hard medians received no cost reduction in this

system. On many roads like two lane rural highways

or smaller city roads, there is no median between the

travel lanes, which was the second best option,

receiving a cost reduction of 10%. While an

emergency vehicle does not have to drive over a curb,

they would have to enter an oncoming traffic lane

which can be dangerous when done improperly. The

best median construction was a soft median or two-

way turn lane in the middle, because they provide

plenty of additional space to move around other

vehicles, and thus this type received the greatest cost

reduction of 20%.

Figure 3: Demonstration of the proposed approach.

After coming up with the adjustment factors, we were

able to demonstrate the proposed route guidance

system to the experts by comparing the standard

shortest distance algorithm to the proposed

emergency route guidance system. As expected,

several routes were the same with either method. This

was the case when there was only one obvious route,

such as from Route 29 and Hydraulic Road to East

Rio Road and Albemarle Square. However, we were

able to find multiple destinations that led to different

routes being suggested. One example of this, as

shown in Figure 3, was when determining the optimal

route from STVFD on Berkmar Dr., between Route

29 and West Rio Road, to Carrsbrook Drive and

Huntington Road. The shortest distance model

suggests taking Northfield Rd. from East Rio Rd. for

a distance of 2.4 miles. That route has several traffic

lights and only one with the pre-emption technology,

while the route the proposed approach suggests

traveling north on Route 29 to Carrsbrook Road is

longer at 2.7 miles, but has larger roads, a faster speed

limit, and several lights with the pre-emption

technology installed. This is the same route the

experts regularly take instead of the shortest distance

route.

4 CONCLUDING REMARKS

This paper proposed and demonstrated a route

guidance approach for emergency vehicles. A Python

based emergency vehicle route guidance algorithm

code was developed and a JavaScript Object Notation

VEHITS 2016 - International Conference on Vehicle Technology and Intelligent Transport Systems

190

(JSON) was used to visualize the network used in this

paper. It is noted that the Python code and JSON can

be used with any network and cost functions as

deemed by emergency vehicle domain experts.

While we could have applied recent advanced

route guidance algorithms such as A* algorithm

(Zeng and Church, 2009), dynamic traffic

assignment-based guidance (Lee and Park, 2008),

reliable route guidance (Nie et al., 2012), the main

purpose of this paper was to demonstrate emergency

vehicle route guidance by utilizing additional road

features. Future research should consider

implementing these advanced algorithms as well as

generalized cost function instead of simple weighted

average used in this paper.

In addition, technology is constantly making

advances, and the field of traffic operations is no

different. Emergency responders, and the fire service

in particular, tend to hold onto tradition sometimes to

their own detriment. There is technology available

that, if adopted, could improve the efficiency of their

responses with very little cost. With even relatively

few additional pieces of information about the road

network in an area, routing can be greatly improved.

For a field in which seconds matter, personnel should

strive for small improvements that could have a large

impact, such as this incremental improvement over

standard routing algorithms to improve speed and

safety in responding to emergencies.

ACKNOWLEDGEMENTS

This research was in part supported by the Global

Research Laboratory Program through the NRF

funded by the Ministry of Science, ICT & Future

Planning (2013K1A1A2A02078326). We are

grateful for anonymous reviewers for their invaluable

comments.

REFERENCES

Bray, Tim. "JSON Redux AKA RFC7159". Ongoing.

Retrieved 25 November 2015.

Data-Driven Documents. n.d. from http://d3js.org.

Dijkstra, E. W. (1959). "A note on two problems in

connexion with graphs". Numerische Mathematik 1:

269–271. doi:10.1007/BF01386390.

Google Maps. n.d. from http://www.maps.google.com.

Lee, J., Park, B., 2008. Evaluation of Route Guidance

Strategies Based on Vehicle-Infrastructure Integration

Under Incident Conditions. Transportation Research

Record: Journal of the Transportation Research Board

2086, 107–114.

Nie, Y., Wu, X., Dillenburg, J.F., Nelson, P.C., 2012.

Reliable route guidance: A case study from Chicago.

Transportation Research Part A: Policy and Practice

46, 403–419.

Zeng, W., Church, R.L., 2009. Finding shortest paths on

real road networks: the case for A*. International

Journal of Geographical Information Science 23, 531–

543.

APPENDIX

1 Python Code

__author__ = 'Danny Brady & B. Brian Park'

import json

MEDIAN_ADJUSTMENT_FACTOR = {}

MEDIAN_ADJUSTMENT_FACTOR['soft'] = 0.8

MEDIAN_ADJUSTMENT_FACTOR['no'] = 0.9

MEDIAN_ADJUSTMENT_FACTOR['hard'] = 1

CONTROL_DEVICE_ADJUSTMENT_FACTOR = {}

CONTROL_DEVICE_ADJUSTMENT_FACTOR['preempti

onSignal'] = 0.6

CONTROL_DEVICE_ADJUSTMENT_FACTOR['signal'] =

0.8

CONTROL_DEVICE_ADJUSTMENT_FACTOR['stop'] = 1

def Dijkstra(graph, source, costFunction):

unvisited_intersections = []

dist = {}

prev = {}

dist[source] = 0

for v in graph['intersections']:

n = v['name']

if n != source:

dist[n] = float("inf")

prev[n] = None

unvisited_intersections.append(v)

while len(unvisited_intersections) > 0:

u = None

current_shotest_distance = float('inf')

for i in unvisited_intersections:

if dist[i['name']] < current_shotest_distance:

u = i

current_shotest_distance = dist[i['name']]

unvisited_intersections.remove(u)

for neighbor_of_u in [road for road in graph['roads'] if

road['source'] == u['name'] or road['target'] == u['name']]:

neighbor_name = neighbor_of_u['target'] if

neighbor_of_u['target'] != u['name'] else neighbor_of_u['source']

neighbor = intersectionFromName(neighbor_name,

graph)

alt = dist[u['name']] + costFunction(u['name'],

Improving Emergency Vehicle Routing with Additional Road Features

191

neighbor['name'], graph)

if alt < dist[neighbor['name']]:

dist[neighbor['name']] = alt

prev[neighbor['name']] = u

return (dist, prev)

def intersectionFromName(name, graph):

return [i for i in graph['intersections'] if i['name'] ==

name][0]

### Cost Functions ###

def linkCostDistance(source, destination, graph):

partial = [road for road in graph['roads'] if road['source']

== source or road['target'] == source]

link = [road for road in partial if road['source'] ==

destination or road['target'] == destination][0]

return float(link['distance'])

def linkCostSpeedWeighted(source, destination, graph):

partial = [road for road in graph['roads'] if road['source']

== source or road['target'] == source]

link = [road for road in partial if road['source'] ==

destination or road['target'] == destination][0]

return float(link['distance']) / float(link['speed'])

def linkCostEmergency(source, destination, graph):

partial = [road for road in graph['roads'] if road['source']

== source or road['target'] == source]

link = [road for road in partial if road['source'] ==

destination or road['target'] == destination][0]

cost = float(link['distance']) / float(link['speed']) * (8 /

float(link['lanes'])) *

MEDIAN_ADJUSTMENT_FACTOR[link['median']]

source_intersection = intersectionFromName(source,

graph)

destination_intersection =

intersectionFromName(destination, graph)

cost = cost *

CONTROL_DEVICE_ADJUSTMENT_FACTOR[source_interse

ction['controlType']] * \

CONTROL_DEVICE_ADJUSTMENT_FACTOR[destination_in

tersection['controlType']]

return cost

if __name__ == "__main__":

g = None

with open("RoadNetwork.json") as json_file:

g = json.load(json_file)

#dist, prev = Dijkstra(g, "Station 8", linkCostDistance)

#dist, prev = Dijkstra(g, "Station 8",

linkCostSpeedWeighted)

dist, prev = Dijkstra(g, "Station 8", linkCostEmergency)

dest_intersection = "Greenbrier Dr @ Whitewood Rd"

#dest_intersection = "Berkmar Dr @ Hilton Heights Rd"

path = []

path.append(intersectionFromName(dest_intersection, g))

while dest_intersection != "Station 8":

p = prev[dest_intersection]

path.append(p)

dest_intersection = p['name']

path.reverse()

for p in path:

print p

2 Complete Road Network Data

{

"intersections":[

{"name":"29 @ Rio Rd

W","controlType":"preemptionSignal"},

{"name":"Berkmar Dr @ Rio Rd

W","controlType":"signal"},

{"name":"29 @ Albemarle

Sq","controlType":"preemptionSignal"},

{"name":"29 @ Woodbrook

Dr","controlType":"preemptionSignal"},

{"name":"29 @ Gander

Dr","controlType":"preemptionSignal"},

{"name":"29 @ Hilton Heights

Rd","controlType":"preemptionSignal"},

{"name":"29 @ Polo Grounds

Rd","controlType":"preemptionSignal"},

{"name":"Four Seasons Dr @ Rio Rd

W","controlType":"signal"},

{"name":"Four Seasons Dr @ Commonwealth

Dr","controlType":"stop"},

{"name":"Commonwealth Dr @ Westfield

Rd","controlType":"stop"},

{"name":"Commonwealth Dr @ Greenbrier

Dr","controlType":"stop"},

{"name":"Earlysville Rd @ Rio Rd

W","controlType":"signal"},

{"name":"Hydraulic Rd @ Whitewood

Rd","controlType":"signal"},

{"name":"Greenbrier Dr @ Whitewood

Rd","controlType":"stop"},

{"name":"Hydraulic Rd @ Georgetown

Rd","controlType":"signal"},

{"name":"Barracks Rd @ Georgetown

Rd","controlType":"signal"},

{"name":"Hydraulic Rd @ Commonwealth

Dr","controlType":"signal"},

{"name":"Hydraulic Rd @ District

Ave","controlType":"signal"},

{"name":"29 @ Hydraulic

Rd","controlType":"signal"},

{"name":"29 @ Fashion Square

Dr","controlType":"preemptionSignal"},

{"name":"29 @ Twentyninth Place

Ct","controlType":"preemptionSignal"},

{"name":"29 @ Branchlands

VEHITS 2016 - International Conference on Vehicle Technology and Intelligent Transport Systems

192

Blvd","controlType":"preemptionSignal"},

{"name":"29 @ Greenbrier

Dr","controlType":"preemptionSignal"},

{"name":"29 @ Seminol

Ct","controlType":"preemptionSignal"},

{"name":"Berkmar Dr @ Woodbrook

Dr","controlType":"signal"},

{"name":"Berkmar Dr @ Hilton Heights

Rd","controlType":"stop"},

{"name":"Rio Rd E @ Albemarle

Square","controlType":"signal"},

{"name":"Rio Rd E @ Old Brook

Rd","controlType":"signal"},

{"name":"Rio Rd E @ Hillsdale

Dr","controlType":"signal"},

{"name":"Rio Rd E @ Huntington

Rd","controlType":"signal"},

{"name":"Rio Rd E @ Greenbrier

Dr","controlType":"signal"},

{"name":"Rio Rd E @ Meadowcreek

Parkway","controlType":"signal"},

{"name":"Rio Rd E @ Melbourne

Rd","controlType":"signal"},

{"name":"Station 8","controlType":"stop"},

{"name":"29 @ Carrsbrook

Dr","controlType":"stop"},

{"name":"Carrsbrook Dr @ Old Brook

Rd","controlType":"stop"},

{"name":"Carrsbrook Dr @ Hillsdale

Dr","controlType":"stop"},

{"name":"Carrsbrook Dr @ Huntington

Rd","controlType":"stop"},

{"name":"Meadowcreek Parkway @

Melbourne Rd","controlType":"stop"},

{"name":"Branchlands Blvd @ Hillsdale

Dr","controlType":"stop"},

{"name":"Hillsdale Dr @ Greenbrier

Dr","controlType":"stop"},

{"name":"29 @ Berkmar

Dr","controlType":"stop"}

"roads":[

{"source":"29 @ Hilton Heights Rd",

"target":"29 @ Polo Grounds Rd", "speed":45,

"distance":2400, "lanes":8, "median":"hard"},

{"source":"29 @ Hilton Heights Rd",

"target":"29 @ Carrsbrook Dr", "speed":45,

"distance":1465, "lanes":8, "median":"hard"},

{"source":"29 @ Gander Dr", "target":"29 @

Carrsbrook Dr", "speed":45, "distance":610, "lanes":8,

"median":"hard"},

{"source":"29 @ Woodbrook Dr", "target":"29

@ Gander Dr", "speed":45, "distance":1650, "lanes":8,

"median":"hard"},

{"source":"29 @ Woodbrook Dr", "target":"29

@ Albemarle Sq", "speed":45, "distance":1685, "lanes":8,

"median":"hard"},

{"source":"29 @ Rio Rd W", "target":"29 @

Albemarle Sq", "speed":45, "distance":685, "lanes":8,

"median":"hard"},

{"source":"29 @ Rio Rd W", "target":"29 @

Fashion Square Dr", "speed":45, "distance":980, "lanes":8,

"median":"hard"},

{"source":"29 @ Berkmar Dr", "target":"29 @

Fashion Square Dr", "speed":45, "distance":520, "lanes":8,

"median":"hard"},

{"source":"29 @ Berkmar Dr", "target":"29 @

Twentyninth Place Ct", "speed":45, "distance":750,

"lanes":8, "median":"hard"},

{"source":"29 @ Twentyninth Place Ct",

"target":"29 @ Branchlands Blvd", "speed":45,

"distance":1210, "lanes":8, "median":"hard"},

{"source":"29 @ Greenbrier Dr", "target":"29

@ Branchlands Blvd", "speed":45, "distance":1750,

"lanes":8, "median":"hard"},

{"source":"29 @ Greenbrier Dr", "target":"29

@ Seminol Ct", "speed":45, "distance":1910, "lanes":8,

"median":"hard"},

{"source":"29 @ Hydraulic Rd", "target":"29

@ Seminol Ct", "speed":45, "distance":1680, "lanes":8,

"median":"hard"},

{"source":"29 @ Rio Rd W",

"target":"Berkmar Dr @ Rio Rd W", "speed":35,

"distance":1485, "lanes":4, "median":"no"},

{"source":"Four Seasons Dr @ Rio Rd W",

"target":"Berkmar Dr @ Rio Rd W", "speed":35,

"distance":3255, "lanes":4, "median":"no"},

{"source":"Four Seasons Dr @ Rio Rd W",

"target":"Earlysville Rd @ Rio Rd W", "speed":35,

"distance":1260, "lanes":4, "median":"no"},

{"source":"Hydraulic Rd @ Whitewood Rd",

"target":"Earlysville Rd @ Rio Rd W", "speed":35,

"distance":3455, "lanes":4, "median":"no"},

{"source":"Hydraulic Rd @ Whitewood Rd",

"target":"Hydraulic Rd @ Georgetown Rd", "speed":35,

"distance":1620, "lanes":4, "median":"no"},

{"source":"Barracks Rd @ Georgetown Rd",

"target":"Hydraulic Rd @ Georgetown Rd", "speed":35,

"distance":4525, "lanes":2, "median":"no"},

{"source":"Hydraulic Rd @ Commonwealth

Dr", "target":"Hydraulic Rd @ Georgetown Rd",

"speed":35, "distance":1815, "lanes":4, "median":"no"},

{"source":"Hydraulic Rd @ Commonwealth

Dr", "target":"Hydraulic Rd @ District Ave", "speed":35,

"distance":950, "lanes":4, "median":"no"},

{"source":"Hydraulic Rd @ Commonwealth

Dr", "target":"Commonwealth Dr @ Greenbrier Dr",

"speed":35, "distance":3165, "lanes":2, "median":"soft"},

{"source":"29 @ Hydraulic Rd",

"target":"Hydraulic Rd @ District Ave", "speed":35,

"distance":1185, "lanes":4, "median":"no"},

{"source":"Berkmar Dr @ Rio Rd W",

"target":"Station 8", "speed":35, "distance":880, "lanes":2,

Improving Emergency Vehicle Routing with Additional Road Features

193

"median":"no"},

{"source":"Station 8", "target":"29 @ Berkmar

Dr", "speed":35, "distance":1920, "lanes":2,

"median":"no"},

{"source":"Berkmar Dr @ Rio Rd W",

"target":"Berkmar Dr @ Woodbrook Dr", "speed":35,

"distance":2180, "lanes":2, "median":"no"},

{"source":"Berkmar Dr @ Hilton Heights Rd",

"target":"Berkmar Dr @ Woodbrook Dr", "speed":35,

"distance":3660, "lanes":2, "median":"no"},

{"source":"29 @ Rio Rd W", "target":"Rio Rd

E @ Albemarle Square", "speed":35, "distance":835,

"lanes":4, "median":"no"},

{"source":"Rio Rd E @ Old Brook Rd",

"target":"Rio Rd E @ Albemarle Square", "speed":35,

"distance":2400, "lanes":4, "median":"no"},

{"source":"Rio Rd E @ Old Brook Rd",

"target":"Rio Rd E @ Hillsdale Dr", "speed":35,

"distance":210, "lanes":4, "median":"no"},

{"source":"Rio Rd E @ Huntington Rd",

"target":"Rio Rd E @ Hillsdale Dr", "speed":35,

"distance":2095, "lanes":4, "median":"soft"},

{"source":"Rio Rd E @ Huntington Rd",

"target":"Rio Rd E @ Greenbrier Dr", "speed":35,

"distance":670, "lanes":4, "median":"soft"},

{"source":"Rio Rd E @ Greenbrier Dr",

"target":"Rio Rd E @ Meadowcreek Parkway",

"speed":35, "distance":2475, "lanes":4, "median":"soft"},

{"source":"Rio Rd E @ Meadowcreek

Parkway", "target":"Rio Rd E @ Melbourne Rd",

"speed":35, "distance":7180, "lanes":2, "median":"no"},

{"source":"Rio Rd E @ Meadowcreek

Parkway", "target":"Meadowcreek Parkway @ Melbourne

Rd", "speed":35, "distance":5330, "lanes":2,

"median":"no"},

{"source":"29 @ Greenbrier Dr",

"target":"Hillsdale Dr @ Greenbrier Dr", "speed":25,

"distance":1215, "lanes":2, "median":"soft"},

{"source":"Branchlands Blvd @ Hillsdale Dr",

"target":"Hillsdale Dr @ Greenbrier Dr", "speed":35,

"distance":1635, "lanes":2, "median":"soft"},

{"source":"Branchlands Blvd @ Hillsdale Dr",

"target":"Rio Rd E @ Hillsdale Dr", "speed":35,

"distance":3335, "lanes":2, "median":"soft"},

{"source":"29 @ Carrsbrook Dr",

"target":"Carrsbrook Dr @ Old Brook Rd", "speed":25,

"distance":5095, "lanes":2, "median":"no"},

{"source":"Carrsbrook Dr @ Hillsdale Dr",

"target":"Carrsbrook Dr @ Old Brook Rd", "speed":25,

"distance":1055, "lanes":2, "median":"no"},

{"source":"Carrsbrook Dr @ Hillsdale Dr",

"target":"Carrsbrook Dr @ Huntington Rd", "speed":25,

"distance":555, "lanes":2, "median":"no"},

{"source":"Rio Rd E @ Old Brook Rd",

"target":"Carrsbrook Dr @ Old Brook Rd", "speed":30,

"distance":6285, "lanes":2, "median":"no"},

{"source":"Rio Rd E @ Hillsdale Dr",

"target":"Carrsbrook Dr @ Hillsdale Dr", "speed":30,

"distance":6020, "lanes":2, "median":"no"},

{"source":"Rio Rd E @ Huntington Rd",

"target":"Carrsbrook Dr @ Huntington Rd", "speed":30,

"distance":7285, "lanes":2, "median":"no"},

{"source":"Commonwealth Dr @ Westfield

Rd", "target":"Commonwealth Dr @ Greenbrier Dr",

"speed":25, "distance":1055, "lanes":2, "median":"no"},

{"source":"Commonwealth Dr @ Westfield

Rd", "target":"Four Seasons Dr @ Commonwealth Dr",

"speed":25, "distance":1545, "lanes":2, "median":"no"},

{"source":"Four Seasons Dr @ Rio Rd W",

"target":"Four Seasons Dr @ Commonwealth Dr",

"speed":25, "distance":3515, "lanes":2, "median":"no"},

{"source":"Greenbrier Dr @ Whitewood Rd",

"target":"Hydraulic Rd @ Whitewood Rd", "speed":25,

"distance":2370, "lanes":2, "median":"no"},

{"source":"Greenbrier Dr @ Whitewood Rd",

"target":"Commonwealth Dr @ Greenbrier Dr",

"speed":25, "distance":510, "lanes":2, "median":"no"},

{"source":"29 @ Greenbrier Dr",

"target":"Commonwealth Dr @ Greenbrier Dr",

"speed":25, "distance":1725, "lanes":2, "median":"soft"},

{"source":"29 @ Woodbrook Dr",

"target":"Berkmar Dr @ Woodbrook Dr", "speed":25,

"distance":1390, "lanes":4, "median":"hard"},

{"source":"29 @ Hilton Heights Rd",

"target":"Berkmar Dr @ Hilton Heights Rd", "speed":25,

"distance":1070, "lanes":4, "median":"hard"}

]

}

VEHITS 2016 - International Conference on Vehicle Technology and Intelligent Transport Systems

194