ScienceDirect

Available online at www.sciencedirect.com

Transportation Research Procedia 00 (2017) 000–000
ScienceDirect www.elsevier.com/locate/procedia
Available online at www.sciencedirect.com
Transportation Research Procedia 00 (2017) 000–000
Available online at www.sciencedirect.com www.elsevier.com/locate/procedia
ScienceDirect
20th EURO Working Group ScienceDirect
Transportation Research Procedia 27 (2017) 69–76
on Transportation Meeting, EWGT 2017, 4-6 September 2017, www.elsevier.com/locate/procedia
Budapest, Hungary
Transportation Research Procedia 00 (2017) 000–000
www.elsevier.com/locate/procedia
20th EURO Working Group on Transportation Meeting, EWGT 2017, 4-6 September 2017,
V2V-Communication, LiDAR System
Budapest, and Positioning Sensors for
Hungary
Future Fusion Algorithms in Connected Vehicles
V2V-Communication, LiDAR System and Positioning Sensors for
Sven Eckelmann a, Toralf Trautmann a, Hagen Ußler b, Benjamin Reichelt b,*,
20th EUROFuture
WorkingFusion
Group onAlgorithms in Connected Vehicles
Transportation Meeting,
Oliver Michler b EWGT 2017, 4-6 September 2017,
Budapest, Hungary
Sven Eckelmann , Toralf Trautmann , Hagen Ußler , Benjamin Reichelt b,*,
a Technik und Wirtschaft Dresden,
a Hochschule für a Friedrich-List-Platz 1,b01069 Dresden, Germany

V2V-Communication, LiDAR OliverSystem

Michler and Positioning Sensors
b Technische Universität Dresden, Hettnerstraße 3,b01062 Dresden, Germany
for
Future
a HochschuleFusion
für Technik undAlgorithms in Connected
Wirtschaft Dresden, Friedrich-List-Platz Vehicles
1, 01069 Dresden,
b Technische Universität Dresden, Hettnerstraße 3, 01062 Dresden, Germany
Germany

Abstract
Sven Eckelmann a, Toralf Trautmann a, Hagen Ußler b, Benjamin Reichelt b,*,
The automotive industry is changing from conventional driving into connected and later on autonomous driving. The fundamental
principle Oliver Michler b
Abstract of this alteration is communication and exchange of data between vehicles and other kind of traffic objects, for example
traffic lights. The knowing aboutfürbasic
a Hochschule Technikconditions fromDresden,
und Wirtschaft all traffic objects within a1,close
Friedrich-List-Platz 01069proximity can ensure a more precise
Dresden, Germany
reaction
The of Advanced
automotive industryDriver Assistant
is changing fromSystems (ADAS).
conventional Thereby
driving into Vehicle-to-Vehicle
connected and later (V2V)
b Technische Universität Dresden, Hettnerstraße 3, 01062 Dresden, Germanyon technology
autonomous contributes
driving. to increase
The fundamental
traffic safety.
principle This
of this paper describes
alteration an investigation
is communication of V2Vofcommunication
and exchange based onand
data between vehicles commercial
other kindOn-Board-Units
of traffic objects,(OBU). These
for example
units,
trafficintegrated
lights. The in knowing
two test-vehicles,
about basictransmit and receive
conditions from data based objects
all traffic on the IEEE
within802.11p
a closestandard (ETSI
proximity canITS-G5).
ensure a Themore messages
precise
include basic
reaction conditions
of Advanced like Assistant
Driver position, motion
Systemsvector and vehicle
(ADAS). Therebyconfiguration parameters.
Vehicle-to-Vehicle (V2V)Those information
technology ensuretoa increase
contributes relative
Abstract
positioning
traffic safety.which
This ispaper
also describes
implemented in the paper on
an investigation of hand. The calculationsbased
V2V communication provide the prerequisites
on commercial for additional
On-Board-Units autonomous
(OBU). These
vehicleintegrated
units, system applications like autonomous
in two test-vehicles, transmit and braking
receiveordata
steering
based maneuver.
on the IEEETo802.11p
improve precision
standard (ETSIandITS-G5).
reliability
Theofmessages
ADAS,
The automotive
statements
include about
basic industry
integrityis
conditions changing
are
like position,from
inevitable. conventional
These
motion vector driving
can beand
achieved
vehicleinto connected
a fusion ofand
byconfiguration later on
different
parameters. autonomous
sensor
Those driving.
information.
information The fundamental
Therefore
ensure athe paper
relative
principle
also presentsof this alteration
possibilities is
to communication
ensure an accurate and exchange
localization of
and data between
object vehicles
detection, and
regarding other
a kind
Light of
positioning which is also implemented in the paper on hand. The calculations provide the prerequisites for additional autonomoustraffic
Detection objects,
and Rangingfor example
(LiDAR)
traffic
Systemlights.
vehicle systemThe
(Velodyne knowing
VLP-16) about
applications and basic conditions
likeanautonomous
Differential from
Global
braking orallsteering
traffic System
Positioning objects (DGPS)
maneuver.within
To aimprove
close proximity
approach. can
andensure
The interaction
precision a more precise
of communication,
reliability of ADAS,
reaction
statements of about
sensor fusion Advanced
algorithms Driver
integrity and Assistant
areintegrity
inevitable. Systems (ADAS).
considerations
These can form
be Thereby
the basis
achieved by Vehicle-to-Vehicle
for autonomous
a fusion (V2V)
sensortechnology
driving.
of different information.contributes
Thereforetotheincrease
paper
traffic safety.possibilities
also presents This paper describes
to ensure anan accurate
investigation of V2Vand
localization communication based
object detection, on commercial
regarding On-Board-Units
a Light Detection (OBU).
and Ranging These
(LiDAR)
units,
© 2017
System integrated in two
The Authors.
(Velodyne test-vehicles,
Published
VLP-16) andby transmitB.V.
an Elsevier
Differential andGlobal
receive data basedSystem
Positioning on the IEEE
(DGPS) 802.11p standard
approach. The (ETSI ITS-G5).
interaction The messages
of communication,
include
Peer-reviewbasicunder
sensor fusion conditions
algorithms like integrity
responsibility
and position, motion vector
of theconsiderations
scientific and vehicle
committee
form the the configuration
ofbasis 20th parameters.
EURO Working
for autonomous Group
driving. Those information ensure
on Transportation a relative
Meeting.
positioning which is also implemented in the paper on hand. The calculations provide the prerequisites for additional autonomous
vehicle
© *2017 system
Corresponding
The applications
author.
Authors. likeby
Tel.: +49
Published autonomous
351 463 36758;
Elsevier B.V. braking
fax: or 463
+49 351 steering
36782.maneuver. To improve precision and reliability of ADAS,
E-mail address:
about benjamin.reichelt@tu-dresden.de
Peer-review under responsibility of the scientific committee of theby20th
statements integrity are inevitable. These can be achieved a fusion
EURO ofWorking
different Group
sensoron information.
TransportationTherefore
Meeting.the paper
also presents possibilities to ensure an accurate localization and object detection, regarding a Light Detection and Ranging (LiDAR)
System (Velodyne
* Corresponding VLP-16)
author. and 351
Tel.: +49 an Differential
463 36758; fax: Global Positioning
+49 351 463 36782. System (DGPS) approach. The interaction of communication,
Keywords:
sensor
E-mailfusion algorithms
address: and integrity considerations form the basis for autonomous driving.
benjamin.reichelt@tu-dresden.de
Autonomous Driving; Connected Vehicle; V2V-Communication; IEEE 802.11p; Relative Positioning; Object Detection; Sensor Fusion

© 2017 The Authors. Published by Elsevier B.V.

Keywords:
Peer-review under responsibility of the scientific committee of the 20th EURO Working Group on Transportation Meeting.
Autonomous
2214-241X © Driving;
2017 TheConnected Vehicle; V2V-Communication;
Authors. Published by Elsevier B.V. IEEE 802.11p; Relative Positioning; Object Detection; Sensor Fusion
Peer-review under responsibility
* Corresponding of the
author. Tel.: +49 351scientific
463 36758;committee
fax: +49of351
the463
20th36782.
EURO Working Group on Transportation Meeting.
E-mail address: benjamin.reichelt@tu-dresden.de
2214-241X © 2017 The Authors. Published by Elsevier B.V.
Peer-review under responsibility of the scientific committee of the 20th EURO Working Group on Transportation Meeting.
Keywords:
Autonomous Driving; Connected Vehicle; V2V-Communication; IEEE 802.11p; Relative Positioning; Object Detection; Sensor Fusion

2214-241X ©
2352-1465 © 2017
2017 The
The Authors.
Authors.Published
Publishedby by
Elsevier B.V.B.V.
Elsevier
Peer-review underresponsibility
Peer-review under responsibility of the
of the scientific
scientific committee
committee of theof20th
theEURO
20th EURO Working
Working Group
Group on on Transportation
Transportation Meeting. Meeting.
10.1016/j.trpro.2017.12.032
70 Sven Eckelmann et al. / Transportation Research Procedia 27 (2017) 69–76
2 S. Eckelmann, T. Trautmann, H. Ußler, B. Reichelt, O. Michler / Transportation Research Procedia 00 (2017) 000–000

1. Introduction

The increasing automation of the driving functions places high demands on the vehicle, the environment and the
driver. In order to navigate track selective, the vehicle needs to know its exact position. Conventional Global
Positioning Systems (GPS) do not reach the required accuracies in the centimeter range. Solutions are based on DGPS
approaches or use software technologies like Software Defined Radio (SDR). Cooperatively, the position can be
determined by the perception of the environment using highly accurate digital maps. The challenge of recognizing
objects in the foveal and peripheral field of view places a high degree of integrity on the sensors and data processing.
The different geometric shapes of objects, their surface properties and their movement behavior require specific
sensors corresponding to the application. Additional transmission characteristics of the atmosphere influence the
accuracy of these sensors in a negative manner. The fusion of the information is the state of the art approach and is
applicable for each individual vehicle. The provision of sensor data goes beyond the vehicle - with ad-hoc wireless
networks such as WLANp or G5.
The paper on hand discusses three different sensor information. The positioning section describes possibilities to
increase the own positioning accuracy. The LiDAR sensor, based on a Velodyne VLP-16, provides highly resoluted
3-D point data and is basicly used for object recognition and tracking. The sensor needs a line of sight to objects which
should be recognized and tracked. But reality, especially urban traffic looks different. Shadowing by other cars and
infrastructure causes a reduction of information. For this reason we are going to aggregate information provided by
the V2V communication.

2. Related Work

There is a multitude of research activities focus on automated and intelligent vehicles and especially on
communication in Intelligent Traffic Systems (ITS). As mentioned before neither communication in ITS nor high
accurate localization by itself can realize connected and autonomous driving. For this reason, the following review
includes both of these subjects.
Because of V2V communication provides a high potential for increasing traffic safety, the current literature deals
especially with the application of these units for ITS. One of the most important aspects when V2V communication
is regarded as a sensor system is the real time ability. Molisch et al. (2009) gives an overview of channel characteristics
including transmission reliability and latency for different environment scenarios. Communication requirements to be
met such as end-to-end latency are listed in Hobert et al. (2015).
Cooperative approaches which exploit the communication between vehicles are proposed in Obst et al. (2012) and
Rohani et al. (2016). Both of them have implemented a DGPS method that increases position accuracy by exchanging
GPS raw measurement between traffic objects. Another relevant work is presented by Alam and Dempster (2013),
where different conventional and modern cooperative positioning systems are discussed.
A basic investigation about the combination of Global Navigation Satellite System (GNSS) positioning and a
LiDAR system can be found in Kim et al. (2015). In this paper, a multi-channel LiDAR sensor is used to build a global
map including curb information of the road based on the logged point clouds. However, the authors already pointed
out to intensify investigations to improve 3-D reconstruction and particularly to enhance the fusion with a cooperative
positioning method. For detection, segmentation and classification of the LiDAR point clouds, an approach is
presented by Babahajiani et al. (2016).

3. Fundamentals

3.1. V2X-Communication and ETSI ITS G5

In the context of Vehicle-to-Anything (V2X) communication, different communication technologies are provided
to cover different requirements concerning flexibility, dynamics, reliability, area coverage, latency, data rate, security
aspects and other performance parameters. An European communication structure is described by the European
Telecommunications and Standards Institute (ETSI) ITS with an associated reference architecture. Based on the IEEE
 Sven Eckelmann et al. / Transportation Research Procedia 27 (2017) 69–76 71
S. Eckelmann, T. Trautmann, H. Ußler, B. Reichelt, O. Michler / Transportation Research Procedia 00 (2017) 000–000 3

802.11p protocol, ETSI ITS G5 is used in Europe as the main standard for ITS operating in the 5 GHz frequency band
(ETSI ES 202 663 (2009)). The frequency spectrum has been allocated especially for vehicular networks. Due to low
latency and ad-hoc communication characteristics, ITS G5 is suitable for road safety and traffic efficiency
applications. The latency of safety-critical messages should not exceed 100 ms (Alam et al. (2016)). Compared to
other short range systems for example LiDAR, the ITS G5 offers an extended view for vehicles by receiving safety
critical data from the environment. Depending on the area of application, a range up to 1000 m can be covered under
ideal conditions (Kloiber et al. (2011)).
Different types of messages are defined by the ETSI ITS protocol stack. In the conducted V2X communication
scenario we transmitted Cooperative Awareness Messages (CAM) and Decentralized Environmental Notification
Messages (DENM). It is designated to use additional messages in future implementations. A safety relevant
application commonly realized in the environment of V2X concerns the distance calculation between traffic
participants. The necessary position information is provided by a GNSS receiver and transmitted via CAM or DENM
messages.

3.2. The LiDAR System

For navigation in highly automated driving it is widely accepted that local instantaneous 3-D maps are needed.
High-resolution results with an optimum of accuracy are provided by LiDAR systems. A LiDAR emits infrared
coherent light from a laser to its environment. The energy is decisive for the classification of the sensor into protective
classes and results from the integral of the pulse over time. Currently produced laser diodes emit a maximum power
of 85 W. The sensitivity of the receiver is defined by the receiver surface and is attenuated by the glare which directly
effects the scattering of infrared light in the receiving branch. The distance to an object is calculated as product of the
half of the runtime and speed of light.
For fixed objects such as cars or buildings, the response follows the normal distribution of the transmit pulse. If the
beams do not cover an object completely, several objects can be received on one channel. This multi-target capability
depends on the manufacturer. The distance to the target, the angular resolution in azimuth (horizontal) and elevation
(vertical) and the size of the objects determine whether an object is classified as a potential collision partner.
For closer objects the additional information about the intensity, generated by the LiDAR, is likely relevant for
filtering and object classification. (Winner et al. (2015))

3.3. Positioning

Conventional GNSS positioning is based on distance measurements calculated by the time of flight of the signal
between the receiver antenna and a satellite. The raw measurements from the observed satellites basically include
pseudorange, carrier phase, Doppler shift and signal strength. Using a least squares algorithm processing all observed
pseudoranges for one epoch, a position estimation can be calculated. Due to signal propagation delay (or advance)
caused by several effects, the measured pseudoranges are corrupted, which directly leads to a position error
respectively position uncertainty (Kaplan and Hegarty (2006)). These errors can be separated basically in common
and non-common errors. Common errors are mainly independent of the receivers as long as they are in a near vicinity
to each other. Examples for this type of error are atmospheric delays, ephemeris errors or satellite clock errors. In
contrast to these errors, non-common errors are dependent on the receiver and its position and include for example
multipath effects and receiver noise. Rohani et al. (2016); Kaplan and Hegarty (2006)
There are different approaches to mitigate the positioning errors. One approach, that is also used for the paper on
hand, is a DGPS approach which can provide decimeter-level accuracy. This method utilizes one GPS receiver on a
well-known position as base station and a second GPS receiver as rover. The base station compares the real measured
observables with its known position and hence calculates correction data, which are transmitted via a data link to the
rover (within a RTCM1 correction message). Kaplan and Hegarty (2006)

1
Radio Technical Commission for Maritime Services.
72 Sven Eckelmann et al. / Transportation Research Procedia 27 (2017) 69–76
4 S. Eckelmann, T. Trautmann, H. Ußler, B. Reichelt, O. Michler / Transportation Research Procedia 00 (2017) 000–000

4. Investigation of different Sensor Systems

4.1. V2X-Technology and Time Delay Correction

For safety reasons, the relative distance between two vehicles 𝑉𝑉1 and 𝑉𝑉2 has to be known for every situation.
Typically, this distance is captured with a radar sensor, which needs a line of sight connection. The only possibility to
figure out the relative distance even for a non-line of sight situation is communication between traffic objects, which
is realized in the following. To calculate the distance between the vehicles, it is necessary to transmit the position from
one participant via IEEE 802.11p radio communication to the others. With the ego position, the receiving vehicle
calculates the euclidean distances in a 3-D local coordinate system (north, east, up). Due to the time delay 𝛿𝛿𝛿𝛿𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝
between positional fix 𝛿𝛿𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝 and the time of distance calculation 𝛿𝛿𝑐𝑐𝑝𝑝𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝 , it is necessary to correct the given
positions of both vehicles. The time delay is calculated in the following way

𝛿𝛿𝛿𝛿𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝 = 𝛿𝛿𝑐𝑐𝑝𝑝𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝 − 𝛿𝛿𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝 . (1)

The positional age is the aggregate of different time components, which can not be quantized in detail. The time
delay comprises delays because of different update intervals from the GNSS and CAM messages, internal processing
time of transmitter and receiver as well as latency due to signal traveling over the wireless interface.
With the information of vehicle speed 𝑣𝑣 and heading 𝛼𝛼, provided by GNSS data, it is possible to calculate the
travelled distance 𝛿𝛿𝛿𝛿𝑥𝑥 in north 𝑁𝑁 and east 𝐸𝐸 direction in the time 𝛿𝛿𝛿𝛿𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝 for each vehicle 𝑥𝑥 as follows

𝛿𝛿𝛿𝛿𝑥𝑥,𝐸𝐸 𝛿𝛿𝛿𝛿𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝,𝑥𝑥 ∗ 𝑣𝑣𝑥𝑥 ∗ sin⁡(𝛼𝛼)

( )=( ). (2)
𝛿𝛿𝛿𝛿𝑥𝑥,𝑁𝑁 𝛿𝛿𝛿𝛿𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝,𝑥𝑥 ∗ 𝑣𝑣𝑥𝑥 ∗ 𝑐𝑐𝑐𝑐𝛿𝛿⁡(𝛼𝛼)

Adding these distances to the original positions in north 𝑛𝑛 and east 𝑒𝑒, the time-corrected (𝛿𝛿𝑐𝑐) position of the
vehicles in each direction can be determined as

𝑒𝑒𝑥𝑥,𝑝𝑝𝑐𝑐 𝑒𝑒𝑥𝑥 + 𝛿𝛿𝛿𝛿𝑥𝑥,𝐸𝐸

(𝑛𝑛 ) = ( ). (3)
𝑥𝑥⁡,𝑝𝑝𝑐𝑐 𝑛𝑛𝑥𝑥 + 𝛿𝛿𝛿𝛿𝑥𝑥,𝑁𝑁

By comparing these corrected positions, the predicted time-corrected distance 𝑑𝑑𝑝𝑝𝑐𝑐 between the vehicles 𝑉𝑉1 and 𝑉𝑉2
can be evaluated at the time of calculation from the differences in east 𝑒𝑒, north 𝑛𝑛 and up 𝑢𝑢:

𝑑𝑑𝑝𝑝𝑐𝑐 = √(𝑒𝑒1,𝑝𝑝𝑐𝑐 − 𝑒𝑒2,𝑝𝑝𝑐𝑐 )2 + (𝑛𝑛1,𝑝𝑝𝑐𝑐 − 𝑛𝑛2,𝑝𝑝𝑐𝑐 )2 + (𝑢𝑢1 − 𝑢𝑢2 )2 . (4)

The described relative positioning with the proposed correction method was validated within an experimental test
scenario with two vehicles, each equipped with a WLANp MK5 OBU module from Cohda Wireless (Cohda (2015)).
In the given scenario in Fig. 1 both vehicles travelled anti-clockwise within an oval testfield with a diametrically
opposed initial position and similar speed. The absolute position of the vehicles was always related to the reference
position (in red) which was surveyed in centimeter-level accuracy before. While travelling around the course, line of
sight and non-line of sight conditions occur through the hills within the oval at its east and west side, where the
distance between both vehicles is also the largest. As the grid in Fig. 1 indicates, the test area was surveyed with a
Leica reference system and therefore the grid can be seen as plausibility check of the calculated distances.
The introduced method for relative distance calculation and correction has proved to be useful in the context of
traffic security. Due to the already existing GNSS inaccuracies, an exact distance is not properly determinable.
Therefore the introduced method can only give an estimated corrected distance. However, it could lead to a safe
condition when vehicles recognize each other in time.
 Sven Eckelmann et al. / Transportation Research Procedia 27 (2017) 69–76 73
S. Eckelmann, T. Trautmann, H. Ußler, B. Reichelt, O. Michler / Transportation Research Procedia 00 (2017) 000–000 5

Fig. 1. Relative positioning using V2V technology.

4.2. LiDAR-System

For our approach we are using a Velodyne VLP16, which can process two targets. With a range of 100 m, a 360°
horizontal field of view and a given angular resolution between 0.1° to 0.4° and a +/- 15° vertical field of view divided
on 16 levels, it generates up to 300,000 points per second with a sampling frequency of 20 Hz (Velodyne (2016)).
For our first approach we are using a simple object classifier and tracker to determine between road surface, static
and dynamic objects. Points, generated from an even surfaced road, are usually arranged in a cocircular order and
have a small deviation. Those points are reconstructed as a driving level with a validity for the observed area. Static
objects can be calculated with the knowledge of the vehicle speed and heading. For dynamic objects in a range up to
20 m we will use an approach according to Kusenbach et al. (2016). Objects in further distance will be recognized by
less points. In Fig. 2 the red car was equipped with the LiDAR. Although the LiDAR is mounted at a height of 1.8 m,
large areas are obscured by objects. So the classification of the orange car is based on points generated by one level
in a distance of approximately 25 m. An object recognition in urban areas will always be a challenge regarding
occlusion and the size of the object which has to be recognized. In purely arithmetical terms, assuming 100 %
reflectivity, good weather conditions and an even road surface, objects with a size of a car, which approach a crossing
area in 100 m, are only recognized by 5 points on one level (Thakur (2016)). A clear classification will base on a
fusion with a complementary sensor system.

Fig. 2. Perception of Velodyne VLP 16 in urban area.

4.3. Positioning

For the positioning part, we use additional hardware next to the already integrated GPS modules on the
communication units. These modules, called NEO-M8P from u-blox, are well suited for high performance positioning
74 Sven Eckelmann et al. / Transportation Research Procedia 27 (2017) 69–76
6 S. Eckelmann, T. Trautmann, H. Ußler, B. Reichelt, O. Michler / Transportation Research Procedia 00 (2017) 000–000

by utilizing a DGPS principle with one module as base station and one as rover. The used modules also support Real
Time Kinematic (RTK) service which provides centimeter-level accuracy. The data exchange between base and rover
is realized by a radio link on the u-blox modules (ublox (2017)).
Before starting, the modules need to be configured correctly. For the investigated usecase, the base station, which
will be the traffic light perspectively, was set up to fixed stationary mode. This position was surveyed in centimeter-
level accuracy with a Leica reference system before. For the data transmission, the RTCM correction messages RTCM
1005, RTCM 1077 and RTCM 1087 need to be enabled. In terms of the rover settings, the RTCM message was set as
input protocol for the radio link. As soon as the rover receive a RTCM message, the correction is processed on the
output data. For reasons of comparability, the real time corrected positions (message UBX-NAV-POSLLH) and the
raw measurements (UBX-RXM-RAWX) from rover and base station have been logged as well. Fig. 3a shows the
rover test track (red line) and the corresponding position of the base station (yellow pin). The track leads through a
highly complex urban environment with buildings and trees next to the road. The surrounding of the test track was
surveyed with the Leica reference system. Plotting the reference positions on the Google Earth Map, a high mapping
accuracy in decimeter-level was figured out. Hence, the map is used for visualization in the following.

a b

Fig. 3. (a) Test trajectory in an highly urban area; (b) Section of test track (green line: SPP, red line: UBX-NAV-POSLLH).

Inspired by Ansari et al. (2013), three position solutions have been calculated respectively recorded for evaluation
of the M8P modules. The first one is a carrier-based kinematic solution, which is calculated from the open source
software RTKlib in the post-process. This solution uses forward and backward RTK calculations and hence can be
regarded as a benchmark solution. However, this solution is not available for real-time processing. The second Single
Point Positioning (SPP) solution is likewise calculated using RTKlib in the post-process. But in contrast, SPP only
processes the rover raw measurements, there is no input in terms of correction data from a base station. The last
evaluated position solution is the real-time available RTK-based POSLLH-data out of the UBX-NAV-protocol of the
M8P modules. If there is a stable radio connection between rover and base, the transmitted RTCM correction messages
are applied on these data. Depending on the resolution of one or more fixed integer carrier phase, the output is a float
or fixed solution. It is noted that, because of the highly urban environment, almost every POSLLH solution during the
measurement was a float solution.
Fig. 3b presents a qualitative and Table 1 a quantitative comparison between the SPP and the UBX-NAV-POSLLH
measurements. In the table, 𝑥𝑥 𝑝𝑝 indicates the average amount-based deviation between the recorded measurement
data and the benchmark solution for latitude respectively longitude. The corresponding standard devation is indicated
as . Fig. 3b as well as Table 1 clearly reveal a large variation and also a pretty high inaccuracy of the SPP values
(green line). In contrast, the red-marked UBX-NAV-POSLLH solution shows accuracy in decimeter-level and a
standard deviation which is a power of ten lower than the one for SPP.
 Sven Eckelmann et al. / Transportation Research Procedia 27 (2017) 69–76 75
S. Eckelmann, T. Trautmann, H. Ußler, B. Reichelt, O. Michler / Transportation Research Procedia 00 (2017) 000–000 7

Table 1. Comparison between different positioning solutions.

SPP vs. Benchmark (deg.) UBX-NAV-POSLLH vs. Benchmark (deg.)

𝑥𝑥 𝑝𝑝 ,𝑐𝑐𝑝𝑝𝑝𝑝 5.0925e-04 6.0440e-06
𝑐𝑐𝑝𝑝𝑝𝑝 2.9683e-04 1.3934e-05
𝑥𝑥 𝑝𝑝 ,𝑐𝑐𝑝𝑝𝑝𝑝 9.7261e-04 1.0996e-05
𝑐𝑐𝑝𝑝𝑝𝑝 9.9071e-04 1.0663e-05

4.4. Vehicle-Set-Up

The LiDAR and the GPS system are mounted on a mobile platform, which can be easily adapted to the roof of
different vehicles (Fig. 4). The antenna of the OBU is attached close to the platform. In our case, we are using a BMW
i3 as experimental car. Additional information like vehicle speed, steering angle and wheel speed are provided by a
CAN interface from the car. Altogether, the information are captured by a mini PC over Ethernet with an unique
timestamp. Thus it is possible to play back data at a later time and to adjust the implemented algorithm.

Fig. 4. Mobile sensor setup at the rooftop of the test vehicle.

5. Conclusion and Future Work

This paper is related to the forthcoming change in the automotive industry from conventional driving to
autonomous driving. Especially the inevitable coexistence of communication technology and highly accurate
positioning was pointed out. After a short introduction into the theoretical basics of V2X communication, LiDAR
technology and GNSS positioning, basic investigations about the above mentioned systems have been conducted and
evaluated.
Using V2X communication, an approach for relative positioning between traffic objects was proposed and realized.
In this context, particularly the time delays have been considered as a crucial component.
For precise object localization, a ° LiDAR system was put into operation and an usual intersection situation
was recorded. For improving the absolute position of the ego vehicle, an investigation with RTK-capable GNSS
modules was conducted. Using a base station to transmit correction data, the accuracy of the rover position could be
increased significantly.
These three individual systems provide high potential for future work. For combining the systems, sensor fusion
has to be conducted in the next steps. Especially the fusion of V2X information with LiDAR data provide high
potential. The goal is to fit the point data provided from the Velodyne with the predefined vehicle dimensions from
the received CAM message defined in ETSI EN 302 637. Based on our results, we are looking forward to extend the
predefined vehicle definition to ensure a better object recognition, especially regarding the heading of the vehicle.
To reference the future fusion architecture, already developed and proved localization methods based on Wireless
Sensor Networks can be performed (Baumbach et al. (2013), Weber et al. (2016)).

References

Alam N. and Dempster A. G., Cooperative positioning for vehicular networks: Facts and future. IEEE Transactions on Intelligent Transportation
Systems, 14(4):1708–1717, Dec 2013.
76 Sven Eckelmann et al. / Transportation Research Procedia 27 (2017) 69–76
8 S. Eckelmann, T. Trautmann, H. Ußler, B. Reichelt, O. Michler / Transportation Research Procedia 00 (2017) 000–000

Alam M., Ferreira J. and Fonseca J., Intelligent transportation systems: Dependable vehicular communications for improved road safety,
volume 52. Springer, 2016.
Ansari K., Wang C., Wang L. and Feng Y., Vehicle-to-vehicle real-time relative positioning using 5.9 ghz dsrc media. In 2013 IEEE 78th Vehicular
Technology Conference (VTC Fall), pages 1–7, Sept 2013.
Babahajiani P., Fan L., Kämäräinen J.-K. and Gabbouj M., Comprehensive automated 3d urban environment modelling using terrestrial laser
scanning point cloud. In 2016 IEEE Conference on Computer Vision and Pattern Recognition Workshops (CVPRW), 2016.
Baumbach R., Weber R. and Michler O., Communication and localization in urban traffic environments using ieee 802.15.4-based wsn. 20th ITS
World Congress Tokyo (ITS World), 2013.
Cohda Wireless Pty Ltd, MK5 V2X On Board Unit, 20151210 MKx OBU, 2015.
ETSI ES. 202 663 v1.1.0. Intelligent Transport Systems (ITS); European profile standard for the physical and medium access control layer of
Intelligent Transport Systems operating in the 5 GHz frequency band, 2009.
Hobert L., Festag A., Llatser I., Altomare L., Visintainer F. and Kovacs A.. Enhancements of v2x communication in support of cooperative
autonomous driving. IEEE Communications Magazine, 53(12):64–70, Dec 2015.
Kaplan E. and Hegarty C., Understanding GPS: principles and applications. Artech house, 2006.
Kusenbach M., Himmelsbach M. and Wuensche H. J., A new geometric 3d lidar feature for model creation and classification of moving objects.
In 2016 IEEE Intelligent Vehicles Symposium (IV), pages 272–278, June 2016.
Kim J. K., Kim J. W., Kim J. H., Jung T. H., Park Y. J., Ko Y. H. and Jung S., Experimental studies of autonomous driving of a vehicle on the road
using lidar and dgps. In 2015 15th International Conference on Control, Automation and Systems (ICCAS), pages 1366–1369, Oct 2015.
Kloiber B., Strang T., Röckl M. and de Ponte-Müller F., Performance of cam based safety applications using its-g5a mac in high dense scenarios.
In 2011 IEEE Intelligent Vehicles Symposium (IV), pages 654–660, June 2011.
Molisch A. F., Tufvesson F., Karedal J. and Mecklenbrauker C. F., A survey on vehicle-to-vehicle propagation channels. IEEE Wireless
Communications, 16(6):12–22, December 2009.
Obst M., Mattern N., Schubert R. and Wanielik G., Car-to-car communication for accurate vehicle localization - the covel approach. In International
Multi-Conference on Systems, Sygnals Devices, pages 1–6, March 2012.
Rohani M., Gingras D. and Gruyer D., A novel approach for improved vehicular positioning using cooperative map matching and dynamic base
station dgps concept. IEEE Transactions on Intelligent Transportation Systems, 17(1):230–239, Jan 2016.
Thakur R., Scanning lidar in advanced driver assistance systems and beyond: Building a road map for next-generation lidar technology. IEEE
Consumer Electronics Magazine, 5(3):48–54, July 2016.
Ublox, NEO-M8P u-blox M8 High Precision GNSS Modules Data Sheet. Document Number UBX-15016656, Revision R05, January 2017.
Velodyne LiDAR, Inc. User‘s Manual and Programming Guide. 63-9243 Rev B Mar 2016.
Weber R. , Balzer P., Michler O. and Mademann E., Improved ro-slam using activity classification for automated v2x infrastructure mapping. In
2016 13th Workshop on Positioning, Navigation and Communications (WPNC), pages 1–8, Oct 2016.
Winner H., Hakuli S. and Wolf G., Handbuch fahrerassistenz-systeme. Technical report, Springer, 2015.