Collective Perception: Impact on Fuel Consumption for Heavy Trucks

Juergen Hauenstein

, Jakob Gromer

, Jan Cedric Mertens

, Frank Diermeyer

and Sven Kraus

Institute of Automotive Technology, TU Munich, Boltzmannstr. 15, Garching, Germany

MAN Truck & Bus SE, Munich, Germany

Keywords: Vehicle-to-Everything, Collective Perception, Cooperative Driving, Heavy Truck.

Abstract: With on-board sensor technology, the environment can only be perceived to a limited extent. This can lead to

energy-inefficient driving maneuvers due to the late perception of objects. The fuel consumption of heavy

trucks is a major cost factor for transport companies, which is why energy-efficient systems are being sought.

With collective perception, perceived objects are exchanged via Vehicle-to-Everything (V2X) and merged to

a common environment model. Therefore, it is possible to achieve a greater awareness, which allows for

improved planning for automated vehicles. In this publication, a system with collective perception and energy-

efficient maneuver planning is presented. The functioning of the collective perception is presented using real

vehicle data. A vehicle simulation shows the positive effect of collective perception in combination with an

energy-efficient maneuver planner for determining the fuel consumption of heavy trucks.

1 INTRODUCTION

A number of publications deal with V2X (Lozano

Domínguez & Mateo Sanguino, 2019). In addition,

there are a large number of projects in the EU dealing

with cooperative driving (Botte et al., 2019). The

hope is that connected and automated vehicles will

make traffic safer in the future (Wang et al., 2020).

One problem with today's vehicles is the limited

visibility their own local sensor technology provides.

First, the range is very limited, e.g., a commercially

available LRR4 radar sensor from Robert Bosch

GmbH has a maximum range of 250 m (Robert Bosch

GmbH, 2014). Second, blind spots can also be caused

by vehicles owing to being in the shade, so that

objects cannot or very late perceived. Messages have

already been designed that use V2X communication,

which allows data to be transmitted directly between

vehicles. The first series-produced vehicles are

already using V2X to exchange warnings, for

example, (ADAC e. V., 2020; Rudschies, 2020).

However, currently, not all new vehicles can

communicate. In addition, automotive companies

advocate that in the future autonomous vehicles must

be able to drive safely even without V2X (Wood et

al., 2019). Therefore, in mixed traffic, vehicles with

and without V2X must be assumed in the future. In

addition to hazard alarms, collective perception is

another way to increase safety using V2X. Here the

captured objects are distributed to other vehicles via

an object list. The sent objects increase the perception

range, because it is now also possible to perceive

hidden objects. The simulation has already shown

that collective perception increases perception and

enhances safety (Günther, 2017).

So far, no experiments with real vehicles and

collective perception have been conducted.

Furthermore, the effect on fuel consumption is

unclear. Fuel consumption is particularly important

for heavy trucks in long-distance haulage, since fuel

costs account for between approximately 30% and

41% of a forwarding agency's total costs. (Esch &

Dahlhaus, 2016; Nowak et al., 2016). In addition,

savings in fuel consumption lead to reduced CO

emissions from internal combustion engines. In view

of the global warming caused by the greenhouse

effect and its negative consequences for people and

the environment (Bundesministerium für Umwelt,

Naturschutz und nukleare Sicherheit [BMU], 2019;

Bunz & Mücke, 2017; Masson-Delmotte, 2018), fuel

consumption has a social relevance in addition to its

economic significance.

350

Hauenstein, J., Gromer, J., Mertens, J., Diermeyer, F. and Kraus, S.

Collective Perception: Impact on Fuel Consumption for Heavy Trucks.

DOI: 10.5220/0010405500002932

In Proceedings of the 7th International Conference on Vehicle Technology and Intelligent Transport Systems (VEHITS 2021), pages 350-361

ISBN: 978-989-758-513-5; ISSN: 2184-495X

2 RELATED WORKS

2.1 Vehicle-to-Everything

Basically, there are two ways to realize V2X.

Messages can either be transmitted via a cellular

network or directly via an ad-hoc network. An

overview of both technologies is given by Weber et

al. (2019), Sjoberg et al. (2017), Festag (2015), Naik

et al. (2019), Ganesan et al. (2020) and Molina-

Masegosa and Gozalvez (2017). Cellular networking

such as 5G NR mode 1 distributes messages via

mobile radio while ad-hoc networks such as IEEE

802.11p usually use WLAN technology. Cellular

networks have the advantage of a theoretically

infinite range and messages can be prioritized, thus

the channel load can be regulated well. In return, ad-

hoc networks have the advantage of operating

independent of the mobile network coverage.

However, the research question regarding which

information is desirable for the realization of driving

functions is initially independent of the transmission

method.

In Europe, the Cooperative Awareness Message

(CAM) (European Telecommunications Standards

Institute, 2014a) and Decentralized Environmental

Notification Message (DENM) (European

Telecommunications Standards Institute, 2014b) are

currently specified. The CAM contains information

about the current vehicle status, such as position or

speed, and is distributed to all surrounding vehicles

via broadcast. The DENM is used for broadcasting

warnings, such as the presence of black ice. The

DENM is only sent when an event occurs and the

position is fixed. Unlike the CAM, the DENM is also

forwarded from one vehicle to another vehicle, using

a multi-hop algorithm to increase the range.

Similar to the standard defined in Europe, the

Basic Safety Message (BSM) exists in the USA

(Kenney, 2011). Here a distinction is made between

Part 1 and Part 2. Part 1 is similar to the CAM and

sends the current vehicle status, while Part 2 contains

information similar to the DENM. In contrast to the

European standard, there is no multi-hop algorithm.

The messages are only distributed by broadcast to

surrounding vehicles.

2.2 Collective Perception and

Cooperative Prediction

Cooperative driving can be divided into several

levels. A distinction can be made between explicit

and implicit communication, as well as according to

the utility. Explicit communication refers to the use

of V2X and is also referred explicit cooperation.

Collective perception is also called cooperative

perception and is the cooperative vehicle function

with the least utility. With cooperative prediction,

driving planes are distributed in the form of

trajectories, which eliminates the need for predicting

other vehicles and can improve planning by reducing

the uncertainty of a wrong prediction. The highest

levels of cooperative driving are negotiating a

common driving strategy and collaborative maneuver

planning (Burger et al., 2017). However, of all the

cooperative levels, collective perception is the only

method that is explicitly designed for mixed traffic

involving V2X and non-V2X road users.

In the Technical Report 103 562 of the European

Telecommunications Standards Institute (2019), it

has been shown that collective perception can

increase awareness. In addition, it has been shown in

various scenarios that collective perception increases

the time-to-collision compared to a local environment

model (Eiermann et al., 2020; Günther, 2017). Allig

and Wanielik (2019), Delooz and Festag (2019 -

2019) and Thandavarayan et al. (2019) presented

possibilities for reducing the channel load, so that

information can be exchanged reliably even during

high traffic density. Currently, the Collective

Perception Service and thus the Collective Perception

Message (CPM) are defined as a standard (European

Telecommunications Standards Institute, 2020a). The

CPM contains all necessary information for creating

a common environment model. If objects are

measured in vehicle coordinates relative to the own

position, then in addition to the object list, a reference

object with the absolute position is also necessary.

In addition to CPM, the Maneuver Coordination

Service and the corresponding Maneuver

Coordination Message (MCM) are currently being

defined as a standard (European Telecommunications

Standards Institute, 2020b). With this message,

cooperative maneuvers can be coordinated via

trajectory exchange, for example, by implementing

the concept of planned and desired trajectories

(Lehmann et al., 2018). However, only by sending the

own plan in the form of a trajectory without of any

other information, it is also possible to represent a

cooperative prediction.

3 SYSTEM OVERVIEW

Figure 1 shows an overview of the overall system.

The local and global perception module, the road

model and road API, the visualization and the COM

Collective Perception: Impact on Fuel Consumption for Heavy Trucks

351

module are implementations of the IMAGinE project.

The IMAGinE project aims at developing cooperative

driving functions (European Center for Information

and Communication Technologies (EICT) GmbH,

2017). The road model defined in IMAGinE

describes the roads in frenet coordinates. Road

information can be accessed through an API by all

modules. Also, a library for handling collision checks

is available. The visualization represents objects and

roads, and is only used for visual monitoring. The

COM module is the software interface to the

communication unit. The desired driving action,

which includes coasting advise and a calculated

trajectory is published by the planning module. The

controller translates the output into control signals,

such as desired acceleration 𝑎



or steering wheel

angle 𝛿



and passes on to the actuators. In the

following, the perception modules and the planner for

the energy-efficient trajectory calculation are

described in more detail.

Figure 1: System overview.

3.1 Cooperative Perception

The vehicle detects objects with its sensor system.

The list of perceived objects is passed on to the local

object fusion. The local perception module assigns

each object to an existing track or creates a new track.

A new track is created if the object state differs too

much from a tracks state. In addition, the objects are

predicted, which is necessary to keep objects that are

not recognized for a short time in the environment

model. The local fusion object list is sent to other

surrounding road users via CPM. Since the object

states are expressed in relative coordinates to the

sender vehicle, the ego-vehicle is also sent as a

reference object, which contains the absolute

position. In addition, all necessary data for the CAM

is determined from the acquired ego-vehicle.

Figure 2: Cooperative trucks form IMAGinE project.

Figure 3: Two cooperative trucks following one non-

cooperative vehicle. Perception from following truck on the

same time and different data types are displayed.

COM‐

Module

LocalSensors

Planner

Local

Perception

Module

Global

Perception

Module

Controller

Actuators

CAM,CPM(Objects)

𝑎



,𝛿



,gear

𝑥



𝑡



,𝑦



𝑡



,drivingaction

CAM(Ego)

𝑥, 𝑦, 𝑣, …

(Ego‐Data)

MCM(Ego)

MCM(Objects)

GlobalObjectList

ObjectList

Road

Model/API

Visualization

ModulesfromIMAGinEProject

CustomModulesandSerialProduct

CPM(Ego)/

LocalObjectList



Ego‐Vehicle

SensorObject

LocalFusionObject

CAMObject

CPMObject

GlobalFusionObject

VEHITS 2021 - 7th International Conference on Vehicle Technology and Intelligent Transport Systems

352

The global perception module uses the ego-data

and the objects from the ego-sensors to provide

objects with absolute positions as input. In addition,

the received objects via CAM and CPM are used as

further input variables. The objects are assigned to

existing tracks or new tracks are created and

additionally predicted into the future in the same way

as in the local fusion. The output is a global object list

which is used by the planner.

Within the IMAGinE project, two cooperative test

vehicles were set up at MAN Truck & Bus SE, which

are shown in Figure 2. Figure 3 shows how collective

perception works in an actual vehicle. Two

cooperative trucks follow a non-cooperative vehicle,

which is detected by the second truck with the local

sensor system. The locally fused object is sent via

V2X to the rear truck. In contrast to the local

environment model, the rear truck perceives two

instead of only one object with the collective

perception.

3.2 Maneuver Planning

Figure 4 gives an overview of the planner. The

planner gets the global object list from the global

perception module as input. For cooperative vehicles,

additional trajectories are received over V2X by the

COM module as MCM, thus the behavior of these

vehicles is known. Non-cooperative vehicles do not

send any information via V2X, therefore, their

driving behavior is predicted in the first step. It is

assumed that they continue to move at constant speed.

Figure 4: Maneuver Planning.

In the second step, possible trajectories for the

ego-vehicle are calculated. The present investigation

does not require complex trajectory calculation, but it

is important that alternative trajectories are

calculated, which also include energy-efficient

trajectories in the form of coasting maneuvers. A

previously defined path defines the strategic decision

of the target and the route. Starting from the current

position, the upcoming path section is linked to a

velocity profile. By defining the velocity to the

position and describing the initial state, a fully

described movement in space over time is given,

which corresponds to the definition of a trajectory

(Biagiotti & Melchiorri, 2009). The velocity profiles

represent either constant accelerations (Equation 1) or

coasting maneuvers with open clutch (Equation 2).

𝑣



𝑡



𝑎∗𝑡𝑣



)

𝑣



𝑡



𝑞



𝑞



∗𝑡𝑞



∗𝑡



𝑞



∗𝑡



)

Here 𝑎 is a constant acceleration. 𝑣 represents

velocity and 𝑡 time. The constants 𝑞



, 𝑞



, 𝑞



and 𝑞



are determined from a coasting test where 𝑞



identical to the corresponding initial velocity 𝑣



the test 80 km/h. The coefficients 𝑞



, 𝑞



and 𝑞



are

dependent on mass. Table 1 shows the driving

maneuvers considered in this paper.

Table 1: Overview of velocity equations and the

corresponding parameter for the considered maneuvers.

Equation 1 describes linear motion, equation 2 describes

non-linear motion (3rd degree polynomial).

Maneuver E

. Parameters

Cruising 1

𝑎0







Acceleration

𝑎1







Slight

Deceleration

𝑎1







Deceleration

𝑎2







Strong

Deceleration

𝑎3







Coasting

(mass 7 t)

𝑞



 0.2355







𝑞



9.3027∗10









𝑞



 2.0612 ∗ 10









Coasting

(mass 40 t)

𝑞



 0.1446







𝑞



2.5459∗10









𝑞



 3.1443 ∗ 10









During the acceleration maneuver, speed is

limited by the maximum allowed speed on the track

section. When maximum speed is reached, the

maneuver continues with constant velocity. Likewise,

during deceleration, the maneuver is limited

COM‐

Module

(MCM)

Local

Sensors

(Ego‐Data)

Trajectory

Calculation

Predictionnon‐

cooperative

vehicles

Global

Perception

Module

(ObjectList)

CostCalculation

Output

(Trajectory,

DrivingAction)

Strategic

Path

Collective Perception: Impact on Fuel Consumption for Heavy Trucks

353

downward by 0 m/s and the velocity is maintained

afterward.

In the following, a collision check is performed.

In our work, a collision is understood as a violation of

the safety distance. According to the German §4

StVO (Straßenverkehrs-Ordnung (StVO), 2017) law,

proper distance must always be maintained so that if

the front vehicle suddenly brakes, the vehicle

following will be able to either decelerate or stop

safely. In addition, the law prescribes further distance

requirements, e.g. that a minimum distance of 50 m

must be maintained for trucks travelling at speeds

above 50 km/h on German freeways, which is not

relevant in the present work, since the initial distances

are sufficiently large. The safety distance can be

defined by the headway. The headway indicates how

much time elapses until two following vehicles reach

the same point on the road. The recommendations of

the countries are not uniform and can be as high as 3

seconds, but for times greater than 2 seconds, a safe

distance can be assumed (Mahmud et al., 2017).

Headway is not calculable during standstill, which is

why a minimum distance 𝑑



with 10 m is defined.

In this paper, the collision is defined as follows:

𝑑max 𝑣∗𝑡



,𝑑





(3)

with

𝑑𝑥



𝑥𝑙



(4)

𝑥



and 𝑙



state the position and length

of the front vehicle. 𝑥 is the position along a lane of

the relevant following vehicle and 𝑑 is the

corresponding distance to the front vehicle. The ITS-

G5 reference point for the position is used, it is

indicated as the front center bumper projected on the

ground (European Telecommunications Standards

Institute, 2014a). 𝑣 is the current speed and 𝑡



is the desired time interval. Derived from the

headway, 2 secs are selected for 𝑡



Table 2: Maneuver costs.

Maneuver Maneuver Costs (Priority)

Cruisin

Acceleration 0

Slight Deceleration 3

Deceleration 4

Strong Deceleration 5

Coastin

Next, the collision free trajectories are charged

with costs. The costs are determined based on the

maneuver, which represents a prioritization of the

maneuvers. Table 2 shows the allocation between

maneuver and costs. The collision-free trajectory with

the lowest costs is set as the output trajectory. This is

passed on to the controller and sent as an MCM to

other road users.

4 SIMULATIVE EVALUATION

4.1 Simulation Environment

The code for the function logic is written in C++ and

integrated in the Framework Robot Operating System

(ROS). Ubuntu 16.04 was used as the operating

system and the corresponding ROS version ROS

Kinetic. By using the ROS Framework, it is possible

to use the same implementation in the simulation as

in the real vehicle. The real-time vehicle simulation

TruckMaker 7.1 has been extended within the

IMAGinE project of IPG. On the one hand, an

interface to ROS was created and on the other hand,

the extension SimNet allows the detailed simulation

of several ego-vehicles (An & Specka, 2019). V2X

communication is simulated by exchanging ROS

topics, which corresponds to communication without

packet loss. A typical tractor-trailer combination used

in long-distance traffic in Europe was selected to act

as the truck. Based on a demo vehicle with a 353 kW

engine available in TruckMaker, a 12-speed

transmission according to Fries (2019) and Wolff

(2016) was added, which is typically used in long-

distance traffic. For the investigations, two trucks

with two different sensor setups are available (Figure

5). The ranges and beam angles for the long-distance

range are based on the radar sensors described by

Baek et al. (2020). The radar sensor for the short

range is specified according to A.D.C. GmbH (2017).

The simulation is based on a perfect sensor model

in which all state variables are known for perceived

objects, e.g. the length of the vehicle, which real radar

sensors cannot measure.

Figure 5: Sensor setups in the semi in top view. Top: only

front sensor. Bottom: round view.

Angle:120°

Range:50m

Angle:20°

Range:200m

VEHITS 2021 - 7th International Conference on Vehicle Technology and Intelligent Transport Systems

354

4.2 Scenarios

The optimal driving strategy can be driven without

traffic. Only when other road users claim driving

space for themselves do they need to adjust their

driving strategy. Through cooperation, these

scenarios can be solved better than without

cooperation. Ulbrich et al. (2015) has classified and

presented various cooperative scenarios. Most of the

scenarios are associated with lane changes or

intersections. In (Rudschies, 2020) the cooperative

scenario, stops are also mentioned. When

investigating the impact on fuel consumption, only

those scenarios are relevant where it can be expected

that using a global environment model will show less

fuel consumption. Improvements regarding fuel

consumption are mainly expected by avoiding

braking followed by an acceleration to the desired

velocity or by performing coasting maneuvers instead

of cruising with an injecting internal consumption

engine. Often these scenarios are equivalent to the

improvement owing to early detection of obstacles or

conflict situations. Based on this, five scenarios were

derived, which are shown in Figure 6. In all scenarios,

cooperative vehicles forward their detected objects

via the CPM and communicate their driving behavior

to other vehicles via the MCM.

In the first scenario, stopping before a traffic jam,

there is a stationary vehicle on the road. Two

cooperative trucks drive toward the vehicle. The first

cooperative vehicle perceives the stationary vehicle

based solely on its own sensors, whereas the second

cooperative can use the information from both CPM

and MCM.

Scenario two, stopping before a traffic jam and

departing, is similar to scenario one. However, in

contrast to scenario one, the first cooperative vehicle

drives past the stationary vehicle and takes a different

route than the last vehicle. In contrast to scenario one,

where the last vehicle could also only react to the first

cooperative vehicle, here the non-communicating

stationary vehicle is most relevant.

In the third scenario, merging before a traffic jam,

a cooperative truck wants to merge into a lane. At the

end of the lane he is merging into, there is a stationary

vehicle. Another cooperative truck is driving in the

same lane as the stationary vehicle and can detect it

earlier than the vehicle that is planning to merge.

The fourth scenario, merging with conflict, is

similar to the third scenario. A cooperative truck

follows a non-cooperative vehicle in a lane. A further

cooperative truck wants to change to the lane of the

non-cooperative vehicle. The lane change is

conflictual, i.e., the merging vehicle is not allowed to

change lanes. Owing to the collective environment,

this conflict is able to be detected earlier.

In the last scenario, turning left with oncoming

traffic, a cooperative truck wants to turn left at an

intersection while the cooperative truck in front

continues straight ahead. The cooperative truck is

approached by an opposing non-cooperative vehicle,

so that this vehicle must first be let through. The

environment model allows earlier detection of the fact

that an immediate turn is not possible.

Figure 6: Schematic representation of investigated

scenarios.

Table 3 shows the examined variations. The

distance, the total mass and the sensor configuration

are varied. The start distance between the two

cooperative vehicles is based on the ranges for

communication. Actual tests have shown a maximum

range of 700 m when using IEEE 802.11p (Almeida

et al., 2018). Mertens et al. (2020) describes 400 m as

the feasible range for trucks using IEEE 802.11p in

direct vision. If the range is less than 200 m, then

Scenario1:Stoppingbeforetrafficjam

Scenario2:Stoppingbeforetrafficjamanddeparting

Scenario3:Mergingbeforetrafficjam

Scenario4:Mergingwithconflict

Scenario5:Turningleftwithoncomingtraffic

Cooperativetruck(ego‐vehicle)

Non‐cooperativevehicle(trafficobjectwithoutV2X)

Cooperativetruck(traffic‐vehiclewithV2X)

Currentvelocity

Path/Routeego‐vehicle

Collective Perception: Impact on Fuel Consumption for Heavy Trucks

355

detection using commercially available radar sensors

is already possible, which is why this is chosen as the

minimum distance for the variations. The total mass

is varied by the use of a semi-trailer, resulting in total

masses of 7 or 40 t. 40 t, which corresponds to the

German §34 StVZO law regulating the maximum

permissible total mass (Straßenverkehrs-Zulassungs-

Ordnung (StVZO), 2017).

For the investigations, the planner is operated with

the local object list from the local perception module

instead of the global object list as the reference

scenario for each variant. This enables a comparison

between global and local environment models.

Table 3: Variations for scenarios.

Scenario Trailer

Vehicle

Distance

Sensor Setup

No,

Yes

200 m, 400 m,

700

Front

No,

Yes

200 m, 400 m,

700

Front

No,

Yes

Front

No,

Yes

- Front, Round

View

No,

Yes

200 m, 400 m,

700

Front

4.3 Results

In section 1, fuel costs are mentioned as the largest

part of the total cost of ownership for freight

forwarders. In addition, the reduction of fuel

consumption is beneficial for the environment. Based

on this, absolute fuel consumption is the most

important evaluation parameter. Only the ego-

vehicle, which benefits from the transmission of the

CPM, is considered in the following. In the scenarios

presented, the other cooperative vehicle only serves

to transmit the sensor data. This vehicle cannot

benefit from the V2X data and drives the same

trajectory both in the scenarios with and without

cooperation, which is why the fuel consumption is

identical in each case and therefore does not need to

be considered.

Figure 7 shows the velocity profiles of the first

scenario with collective perception as a function of

distance, and the velocity profile when using the local

environment model is also shown. It is recognizable

that the farther the distance or the communication

range is, the earlier a coasting maneuver can be

initiated and the longer the coasting maneuver is.

Figure 8 shows the fuel consumption. The longer the

coasting maneuver is, the greater the benefit of the

environment model. Also visible, the larger the mass

is, the greater the reduction in fuel consumption.

In scenario two, the velocity profiles (Figure 9)

show a different behavior, although the reason for the

necessary deceleration, a stationary vehicle, is the

same. The front cooperative vehicle leaves the lane

and therefore only the stationary vehicle is relevant

for the collision check. The cooperative vehicle sends

a CPM as long as the stationary vehicle is detected by

its own sensors. After the lane change, the object is

no longer within the range of vision of the sensors and

is therefore no longer detected. Objects are further

predicted over a horizon of 2 seconds after the last

CPM has been received. If the distance between

objects is great, a roll maneuver is initiated and

continued until the object is removed from the global

environment model owing to a lack of new

information. The vehicle then accelerates to the

desired speed again and brakes only when the vehicle

is perceived by its own sensors. Consequently, fuel

consumption at the distances 400 and 700 m is much

higher than at 200 m where acceleration to the desired

speed is not necessary (Figure 10).

In the scenario merging in front of the beginning

of a traffic jam, the velocity profiles (Figure 11) show

that with collective perception, a coasting maneuver

is performed before the necessary braking is

performed. Reducing fuel consumption also depends

on the total mass (Figure 12).

The velocity profiles in scenario four, merging

with conflict, show that a conflict cannot be avoided

with a local environment model and only front

sensors, since the speed is not adjusted here (Figure

13). With extended local environment sensors, the

vehicle brakes to 5 m/s, whereas with collective

perception less speed reduction is necessary. Again,

Figure 14 shows a significant improvement in fuel

consumption when using collective perception.

In the last scenario, turning left with oncoming

traffic, the velocity profiles (Figure 15) show similar

behavior with and without collective perception. In

all variants, braking must be applied up to the

maximum permissible curve speed. Braking cannot

be avoided with collective perception. Avoiding

braking to a standstill has little effect on fuel

consumption. Likewise, higher communication

ranges or greater distances have no effect on fuel

consumption (Figure 16). Differences in fuel

consumption are mainly due to the non-deterministic

behavior of ROS.

VEHITS 2021 - 7th International Conference on Vehicle Technology and Intelligent Transport Systems

356

Figure 7: Velocity profiles of the first scenario, stopping

behind traffic jam.

Figure 8: Fuel consumption of the first scenario, stopping

behind traffic jam.

Figure 9: Velocity profiles of the second scenario, stopping

before traffic jam and departing.

Figure 10: Fuel consumption of the second scenario,

stopping before traffic jam and departing.

Figure 11: Velocity profiles of the third scenario, merging

before traffic jam.

Figure 12: Fuel consumption of the third scenario, merging

before traffic jam.

Collective Perception: Impact on Fuel Consumption for Heavy Trucks

357

Figure 13: Velocity profiles of the fourth scenario, merging

with conflict.

Figure 14: Fuel consumption of the fourth scenario,

merging with conflict. Sensor setup round view.

Figure 15: Velocity profiles of the fifth scenario, turning

left with oncoming traffic.

Figure 16: Fuel consumption of the fifth scenario, turning

left with oncoming traffic.

5 DISCUSSION

The benefit of the global environment model in

individual situations was shown in the previous

section. However, the benefit for forwarding agents

depends on the frequency of the situations. Therefore,

a detailed analysis for the occurrence of situations in

the daily routine of long-distance haulage is

necessary. Problematic here is the complex data

acquisition of sensor values, which is however

necessary, so that these can be used for the global

environment model and thus the analysis in the

follow-up. In addition, storing large amounts of data

is a challenge.

Various parameters can be set for the environment

model, e.g. the lifetime of the objects if no sensor

detects them anymore. If objects are discarded at an

early stage, it is possible that a vehicle can stop

driving in a fuel-efficient manner as in scenario 2.

Predicting objects for a very long time can lead to

unrealistic driving behavior. In (Schubert et al.,

2008), for example, the assumption that vehicles

continue to move with constant acceleration is

considered a valid approach, but in practice it may be

that the object can leave the road or the driving lane

if the prediction is long, especially when cornering

(Figure 17). More complex models may be able to

predict objects better, e.g. by map matching onto the

road. However, under certain circumstances a large

number of objects must be predicted, which can lead

to computational time problems.

VEHITS 2021 - 7th International Conference on Vehicle Technology and Intelligent Transport Systems

358

Figure 17: Long prediction with object leaving road.

The range of the communication has little

meaning for the benefit of the collective perception if

the distance to the received object in the CPM is

significantly larger than the sensor range and at the

same time the prediction time of the objects is very

short. This is particularly clear in scenario 2, where a

meaningful coasting process is interrupted again after

the end of the prediction period.

Despite higher awareness, the collective

perception cannot lead to lower fuel consumption in

every scenario. For example, when turning left in the

face of oncoming traffic, it is necessary to decelerate

to the maximum possible curve speed, which is close

to zero anyway.

6 CONCLUSIONS

In the presented paper, the benefits of collective

perception on fuel consumption were shown. A

system architecture with collective perception,

cooperative prediction and a maneuver planner that

allows energy-efficient driving was presented. The

correct functioning of the collective perception was

proven by simulations and real-world tests. In the

simulative evaluation, up to 0.526 l fuel could be

saved in individual situations. The analysis showed

that especially heavy trucks with high mass benefit

from the collective perception. In addition, it was

shown that a long prediction after which objects are

no longer perceived is a decisive factor for saving

fuel.

In future work, the simulative results regarding

fuel consumption with collective perception will be

confirmed in actual tests. Two cooperative trucks and

a test track are available for this purpose. According

to Burger et al. (2017), negotiated and collaborative

maneuver planning offer the greatest utility in

cooperative driving. Therefore, in the following

investigations, cooperative maneuver planning with

trajectories and their effect on fuel consumption will

be investigated in simulation as well as in real-world

tests.

ACKNOWLEDGMENTS

This work is supported by MAN Truck & Bus SE and

the IMAGinE (Intelligent Maneuver Automation -

cooperative hazard avoidance in realtime) project.

IMAGinE is founded by the German Federal Ministry

for Economic Affairs and Energy (BMWi).

CONTRIBUTIONS

As first and corresponding author, Juergen

Hauenstein initiated the paper, wrote the original

draft, defined the approach and contributed the main

parts of the conceptualization, methodology and

investigation goals. Jakob Gromer integrated the

perception modules, developed the planner module

with interfaces and evaluated the concept as a part of

his master thesis. Jakob Gromer contributed to the

conceptualization, methodology and investigation.

Jan Cedric Mertens contributed to driving tests

related to the collective perception and supported the

methodology with discussions and insights. Frank

Diermeyer and Sven Kraus contributed to the concept

of the research project and revised the paper for

intellectual content. Frank Diermeyer and Sven Kraus

gave final approval for the version to be published

and agree to all aspects of the work. As a guarantor,

Frank Diermeyer accepts responsibility for the

overall integrity of the paper.

REFERENCES

ADAC e. V. (2020). Welche Hersteller bieten bereits C2X

an? ADAC-Umfrage 6/2020. https://assets.adac.de/im

age/upload/v1595919606/ADAC-eV/KOR/Text/PDF/

Umfrage_Hersteller_Car2X_dl45xm.pdf.

Allig, C., & Wanielik, G. (2019). Dynamic Dissemination

Method for Collective Perception. In IEEE ITSC 2019:

2019 22st International Conf. on Intelligent

Transportation Systems (ITSC) Auckland, New Zealand,

October 27-30, 2019 (pp. 3756–3762). IEEE.

Almeida, T. T., C. Gomes, L. de, Ortiz, F. M., Junior, J. G.

R., & Costa, L. H. M. K. (2018). IEEE 802.11p

Performance Evaluation: Simulations vs. Real

Experiments. In IEEE ITSC 2018: 2018 21st Intl. Conf.

on Intelligent Transportation Systems (ITSC) Maui,

Hawaii, USA, November 4-7, 2018 (pp. 3840–3845).

IEEE. https://doi.org/10.1109/I TSC.2018.8569676.

Ego‐Vehicle CAMObject GlobalFusionObject

Collective Perception: Impact on Fuel Consumption for Heavy Trucks

359

An, N., & Specka, F. (2019). Entwicklung und Test

kooperativer Fahrfunktionen in virtueller Umgebung.

In Aaet Automatisiertes und Vernetztes Fahren:

Beiträge zum gleichnamigen 20. Braunschweiger

Symposium am 6. Und 7. Februar 2019, Stadthalle,

Braunschweig (1st ed.). ITS mobility e.V.

Baek, M., Jeong, D., Choi, D., & Lee, S. (2020). Vehicle

Trajectory Prediction and Collision Warning via Fusion

of Multisensors and Wireless Vehicular

Communications. Sensors (Basel, Switzerland), 20(1).

https://doi.org/10.3390/s20010288.

Biagiotti, L., & Melchiorri, C. (2009). Trajectory Planning

for Automatic Machines and Robots. Springer Berlin

Heidelberg. https://doi.org/10.1007/978-3-540-85629-0.

Botte, M., Pariota, L., D’Acierno, L., & Bifulco, G. N.

(2019). An Overview of Cooperative Driving in the

European Union: Policies and Practices. Electronics,

8(6), 1–25. https://doi.org/10.3390/electronics8060616.

Bundesministerium für Umwelt, Naturschutz und nukleare

Sicherheit (BMU). (2019). Klimaschutz in Zahlen:

Fakten, Trends und Impulse deutscher Klimapolitik.

https://www.bmu.de/fileadmin/Daten_BMU/Pools/Bros

chueren/klimaschutz_zahlen_2019_broschuere_bf.pdf.

Bunz, M., & Mücke, H.-G. (2017). Klimawandel –

physische und psychische Folgen [Climate change -

physical and mental consequences]. Bundes-

gesundheitsblatt, Gesundheitsforschung, Gesundheitss

chutz, 60(6), 632–639. https://doi.org/10.1007/s00103-

017-2548-3.

Burger, C., Orzechowski, P. F., Tas, O. S., & Stiller, C.

(2017). Rating cooperative driving: A scheme for

behavior assessment. In Ieee ITSC 2017: 20th

International Conference on Intelligent Transportation

Systems : Mielparque Yokohama in Yokohama,

Kanagawa, Japan, October 16-19, 2017 (pp. 1–6).

IEEE. https://doi.org/10.1109/ITSC.2017.8317794.

Delooz, Q., & Festag, A. (2019, November - 2019,

November). Network Load Adaptation for Collective

Perception in V2X Communications. In 2019 IEEE

International Conference on Connected Vehicles and

Expo (ICCVE) (pp. 1–6). IEEE. https://doi.org/

10.1109/ICCVE45908.2019.8964988.

Eiermann, L., Wirthmüller, F., Massow, K., Breuel, G., &

Radusch, I. (2020). Driver Assistance for Safe and

Comfortable On-Ramp Merging Using Environment

Models Extended through V2X Communication and

Role- Based Behavior Predictions. In 2020 IEEE 16th

International Conference on Intelligent Computer

Communication and Processing (ICCP 2020). IEEE.

Esch, T., & Dahlhaus, U. (2016). Antrieb. In E. Hoepke &

S. Breuer (Eds.), ATZ / MTZ-Fachbuch.

Nutzfahrzeugtechnik: Grundlagen, Systeme,

Komponenten (8th ed., pp. 403–540). Springer Vieweg.

European Center for Information and Communication

Technologies (EICT) GmbH. (2017). IMAGinE

Intelligente Manöver Automatisierung - kooperative

Gefahrenvermeidung in Echtzeit: Lösungen für

kooperatives Fahren. https://imagine-online.de/erge

bnisse-publikationen/

European Telecommunications Standards Institute (2014a-

11). Intelligent Transport Systems (ITS); Vehicular

Communications; Basic Set of Applications; Part 2:

Specification of Cooperative Awareness Basic Service

(ETSI EN 302 637-2). http://www.etsi.org.

European Telecommunications Standards Institute (2014b-

11). Intelligent Transport Systems (ITS); Vehicular

Communications; Basic Set of Applications; Part 3:

Specifications of Decentralized Environmental Noti-

fication Basic Service (ETSI EN 302 637-3). http://

www.etsi.org.

European Telecommunications Standards Institute (2019-

12). Intelligent Transport Systems (ITS); Vehicular

Communications; Basic Set of Applications; Analysis of

the Collective Perception Service (CPS); (Technical

Report ETSI TR 103 562). http://www.etsi.org.

European Telecommunications Standards Institute. (2020a,

November 4). Work Programme: Details of 'DTS/ITS-

00167' Work Item. https://portal.etsi.org/webapp/

WorkProgram/Report_WorkItem.asp?wki_id=46541.

European Telecommunications Standards Institute. (2020b,

November 4). Work Programme: Details of 'DTS/ITS-

00184' Work Item. https://portal.etsi.org/webapp/

WorkProgram/Report_WorkItem.asp?WKI_ID=53496

&curItemNr=1&totalNrItems=1&optDisplay=10&qSO

RT=HIGHVERSION&qETSI_ALL=&SearchPage=TR

UE&qETSI_NUMBER=103+561&qINCLUDE_SUB_

TB=True&qINCLUDE_MOVED_ON=&qSTOP_FLG

=&qKEYWORD_BOOLEAN=&qCLUSTER_BOOLE

AN=&qFREQUENCIES_BOOLEAN=&qSTOPPING

_OUTDATED=&butSimple=Search&includeNonActiv

eTB=FALSE&includeSubProjectCode=&qREPORT_T

YPE=SUMMARY.

Festag, A. (2015). Standards for vehicular communi-cation

— from IEEE 802.11p to 5G. E & I Elektrotechnik Und

Informationstechnik, 132(7), 409–416. https://doi.org/

10.1007/s00502-015-0343-0.

Fries, M. J. (2019). Maschinelle Optimierung der

Antriebsauslegung zur Reduktion von CO2-Emissionen

und Kosten im Nutzfahrzeug [Dissertation]. Technische

Universität München, München.

Ganesan, K., Lohr, J., Mallick, P. B., Kunz, A., &

Kuchibhotla, R. (2020). NR Sidelink Design Overview

for Advanced V2X Service. IEEE Internet of Things

Magazine, 3(1), 26–30.

https://doi.org/10.1109/IOTM.0001.1900071.

A.D.C. GmbH. (2017). SRR 20X /-2 /-2C /-21 Short Range

Radar. https://www.continental-automotive.com/geta

ttachment/e98ad36c-49d6-400f-8b99-32ff1ef4e45b/S

RR20X-Datasheet-EN.pdf.pdf.

Günther, H.-J. (2017). Collective Perception in Vehicular

Ad-hoc Networks [Dissertation]. Technische

Universität Braunschweig, Braunschweig.

Kenney, J. B. (2011). Dedicated Short-Range

Communications (DSRC) Standards in the United

States. Proceedings of the IEEE, 99(7), 1162–1182.

https://doi.org/10.1109/JPROC.2011.2132790.

Lehmann, B., Gunther, H.-J., & Wolf, L. (2018). A Generic

Approach towards Maneuver Coordination for

Automated Vehicles. In IEEE ITSC 2018: 2018 21st

VEHITS 2021 - 7th International Conference on Vehicle Technology and Intelligent Transport Systems

360

International Conference on Intelligent Transportation

Systems (ITSC) Maui, Hawaii, USA, November 4-7,

2018 (pp. 3333–3339). IEEE. https://doi.org/10.

1109/ITSC.2018.8569442.

Lozano Domínguez, J. M., & Mateo Sanguino, T. J. (2019).

Review on V2X, I2X, and P2X Communications and

Their Applications: A Comprehensive Analysis over

Time. Sensors (Basel, Switzerland), 19(12).

https://doi.org/10.3390/s19122756.

Mahmud, S. M. S., Ferreira, L., Hoque, M. S., & Tavassoli,

A. (2017). Application of proximal surrogate indicators

for safety evaluation: A review of recent developments

and research needs. IATSS Research, 41(4), 153–163.

https://doi.org/10.1016/j.iatssr.2017.02.001.

Masson-Delmotte, V. (Ed.). (2018). Global Warming of 1.5

ºC: An IPCC Special Report on the impacts of global

warming of 1.5°c above pre-industrial levels and

related global greenhouse gas emission pathways, in

the context of strengthening the global response to the

threat of climate change, sustainable development, and

efforts to eradicate poverty. Intergovernmental Panel

on Climate Change (IPCC). https://www.ipcc.ch/sr15/

Mertens, J. C., Erb, D., Kraus, S., & Diermeyer, F. (2020).

ITS-G5 Antenna Position on Trucks. In 2020 IEEE

Intelligent Vehicles Symposium (IV). IEEE.

Molina-Masegosa, R., & Gozalvez, J. (2017). LTE-V for

Sidelink 5G V2X Vehicular Communications: A New

5G Technology for Short-Range Vehicle-to-Everything

Communications. IEEE Vehicular Technology

Magazine, 12(4), 30–39. https://doi.org/ 10.1109/MVT.

2017.2752798.

Naik, G., Choudhury, B., & Park, J.-M. (2019). IEEE

802.11bd & 5G NR V2X: Evolution of Radio Access

Technologies for V2X Communications. IEEE Access,

7, 70169–70184. https://doi.org/10.1109/ACCESS.2

019.2919489.

Nowak, G., Maluck, J., Stürmer, C., & Pasemann, J. (2016).

The era of digitized trucking: Transforming the

logistics value chain. PwC Strategy& Germany. https://

www.strategyand.pwc.com/media/file/The-era-of-digit

ized-trucking.pdf.

Robert Bosch GmbH. (2014). Chassis Systems Control:

Fourth generation long-range radar sensor (LRR4).

http://cds.bosch.us/themes/bosch_cross/amc_pdfs/LR

R4_292000P0ZH_EN_low.pdf.

Rudschies, W. (Ed.). (2020). C2X im VW Golf 8: Erster

ADAC Test. ADAC e. V. https://www.adac.de/rund-

ums-fahrzeug/tests/assistenzsysteme/c2x-im-vw-golf-8/

Schubert, R., Richter, E., & Wanielik, G. (2008).

Comparison and evaluation of advanced motion models

for vehicle tracking. In 2008 IEEE 11th International

Conf. on Information Fusion (pp. 730–735). IEEE.

Sjoberg, K., Andres, P., Buburuzan, T., & Brakemeier, A.

(2017). Cooperative Intelligent Transport Systems in

Europe: Current Deployment Status and Outlook. IEEE

Vehicular Technology Magazine, 12(2), 89–97. https://

doi.org/10.1109/MVT.2017.2670018.

Straßenverkehrs-Ordnung (StVO), May 18, 2017. https://

www.gesetze-im-internet.de/stvo_2013/

Straßenverkehrs-Zulassungs-Ordnung (StVZO), May 18,

2017. https://www.gesetze-im-internet.de/stvzo_2012/

Thandavarayan, G., Sepulcre, M., & Gozalvez, J. (2019).

Analysis of Message Generation Rules for Collective

Perception in Connected and Automated Driving. In

IEEE IV 2019: IEEE Intelligent Vehicles Symposium

(IV) Paris, France, June 09-12, 2019 (pp. 132–137).

IEEE.

Ulbrich, S., Grossjohann, S., Appelt, C., Homeier, K.,

Rieken, J., & Maurer, M. (2015). Structuring

Cooperative Behavior Planning Implementations for

Automated Driving. In 2015 IEEE 18th International

Conf. on Intelligent Transportation Systems (ITSC

2015): Las Palmas], Gran Canaria, Spain, 15 - 18

September 2015 (pp. 2159–2165). IEEE. https://doi.org

/10.1109/ITSC.2015.349.

Wang, Z., Bian, Y., Shladover, S. E., Wu, G., Li, S. E., &

Barth, M. J. (2020). A Survey on Cooperative

Longitudinal Motion Control of Multiple Connected

and Automated Vehicles. IEEE Intelligent Trans-

portation Systems Magazine, 12(1), 4–24. https://

doi.org/10.1109/MITS.2019.2953562.

Weber, R., Misener, J., & Park, V. (2019). C-V2X - A

Communication Technology for Cooperative,

Connected and Automated Mobility. In Mobile

Communication - Technologies and Applications; 24.

ITG-Symposium (pp. 111–116). VDE.

Wolff, S. (2016). Mehrzieloptimierung von schweren

Nutzfahrzeuggetrieben zur Verbesserung der

Transporteffizienz und der TCO [Semesterarbeit].

Technische Universität München, München.

Wood, M., Robbel, P., Maass, M., Tebbens, R. D., Meijs,

M., Harb, M., Reach, J., Robinson, K., Wittmann, D.,

Srivastava, T., Bouzouraa, M. E., Liu, S., Wang, Y.,

Knobel, C., Boymanns, D., Löhning, M., Dehlink, B.,

Kaule, D., Krüger, R.,. Schlicht, P. (2019). Safety First

for Automated Driving. Aptiv Services US, LLC; Audi

AG; Bayrische Motoren Werke AG; Beijing Baidu

Netcom Science Technology Co., Ltd; Continental Teves

AG & Co oHG; Daimler AG; FCA US LLC; HERE

Global B.V.; Infineon Technologies AG; Intel;

Volkswagen AG. https://www.daimler.com/dokumente

/innovation/sonstiges/safety-first-for-automated-driving.

pdf.

Collective Perception: Impact on Fuel Consumption for Heavy Trucks

361