A Survey on Decentralized Cooperative Maneuver Coordination

for Connected and Automated Vehicles

Daniel Maksimovski

, Andreas Festag

∗ b

and Christian Facchi

Technische Hochschule Ingolstadt/ CARISSMA, Esplande 10, Ingolstadt, Germany

Keywords:

V2X Communications, Cooperative Driving, Maneuver Coordination, Automated Vehicle.

Abstract:

V2X communications can be applied for maneuver coordination of automated vehicles, where the vehicles

exchange messages to inform each other of their driving intentions and to negotiate for joint maneuvers.

For motion and maneuver planning of automated vehicles, the cooperative maneuver coordination extends

the perception range of the sensors, enhances the planning horizon and allows complex interactions among

the vehicles. For speciﬁc scenarios, various schemes for maneuver coordination of connected automated

vehicles exist. Recently, several proposals for maneuver coordination have been made that address generic

instead of speciﬁc scenarios and apply different schemes for the message exchange of driving intentions and

maneuver negotiation. This paper presents use cases for maneuver coordination and classiﬁes existing generic

approaches for decentralized maneuver coordination considering implicit and explicit trajectory broadcast,

cost values and space-time reservation. We systematically describe the approaches, compare them and derive

future research topics.

1 INTRODUCTION

Automated and self-driving vehicles have the poten-

tial to reshape the automotive industry and mobility

by improving the trafﬁc safety and efﬁciency. In the

last two decades, the research and development of ad-

vanced driver-assistance systems (ADAS) and auto-

mated driving functions have seen a huge increase,

both in industry and academia.

Automated driving relies on on-board sensors that

perceive the environment. Their limitations can be

overcome by Vehicle-to-Everything (V2X) commu-

nications. V2X communication offers the possibil-

ity to extend the perception range and enhance the

sensing of the vehicles by having a better represen-

tation of the environment. By exchanging driving

intentions among vehicles, planning algorithms can

enlarge their planning horizon and rely on direct in-

formation from other vehicles instead of predicting

their behavior based on local sensor data. Finally,

V2X communication allows for maneuver negotia-

tion based on a bidirectional message exchange po-

https://orcid.org/0000-0002-3414-1069

https://orcid.org/0000-0001-5254-6425

https://orcid.org/0000-0002-7762-9419

∗

Also with Fraunhofer Application Center for Connected

Mobility and Infrastructure, Ingolstadt, Germany.

tentially with complex interactions.

V2X communication comprises the communica-

tion among vehicles, with pedestrians, the roadside

infrastructure and networks. After several years of

research, development and standardization, two ac-

cess technologies are available for safety and trafﬁc

efﬁciency applications, i.e., WLAN-V2X (or ITS-G5

in Europe) and Cellular-V2X (Sj

oberg et al., 2017;

Molina-Masegosa and Gozalvez, 2017). Both oper-

ate in the 5.9 GHz frequency band allocated for road

safety and trafﬁc efﬁciency applications and enable a

direct ad hoc communication among vehicles applica-

ble for maneuver coordination.

The deployment of the V2X communications can

be divided in three subsequent phases, in which ap-

plications with an increasing level of complexity and

communication requirements are (or will be) imple-

mented. In the ﬁrst phase, “Day-1” applications

exchange vehicle state information (position, speed,

etc.) and share the occurrence of dangerous situa-

tions. They primarily cover applications for driver in-

formation and warning. In the European V2X com-

munication system, these applications use the peri-

odically broadcast Cooperative Awareness Message

(CAM) (ETSI EN 302 637-2)

and the event-driven

Decentralized Environmental Notiﬁcation Message

ETSI standards are available at http://etsi.org.

100

Maksimovski, D., Festag, A. and Facchi, C.

A Survey on Decentralized Cooperative Maneuver Coordination for Connected and Automated Vehicles.

DOI: 10.5220/0010442501000111

In Proceedings of the 7th International Conference on Vehicle Technology and Intelligent Transport Systems (VEHITS 2021), pages 100-111

ISBN: 978-989-758-513-5

Figure 1: Classiﬁcation of approaches for decentralized generic maneuver coordination.

(DENM) (ETSI EN 302 637-3). In the second phase,

the “Day-2” applications rely on sharing of sensor

data; more precisely on the exchange of detected ob-

jects in a vehicle’s vicinity with the Collective Per-

ception Messages (CPM), which is currently being

standardized in ETSI (ETSI TR 102 562). In the

third phase, the main emphasis is on the coopera-

tive maneuver coordination between Connected and

Automated Vehicles (CAVs). This coordination ex-

tends the other communication services by dedicated

messages for the communication of the vehicle’s ma-

neuver intentions. While being a research topic, ma-

neuver coordination is already considered in the early

standardization process (e.g., draft ETSI TR 103 578).

Cooperative maneuver coordination aims at utiliz-

ing V2X communication for the coordination of ma-

neuvers among vehicles in order to achieve safe and

efﬁcient driving. During this process, the involved

CAVs exchange maneuver intentions that inﬂuence

their driving behavior and agree on cooperative joint

maneuvers based on each other needs in a negotiation

phase with a deﬁned number of coordination mes-

sages. The sharing of the intentions and coordination

of maneuvers among the CAVs is expected to enhance

the automated driving by avoiding conﬂicts or colli-

sion risks, and bring the deployment of fully driver-

less vehicles closer to reality.

Existing coordination approaches can mainly be

categorized in two ways. The ﬁrst one separates

them into centralized and decentralized approaches.

Centralized approaches have a central system that re-

ceives all the information and communicates the ma-

neuvers with the respective vehicles. The decentral-

ized approach does not consider a global planner en-

tity, but is solely based on communication and coor-

dination among the involved participants which can

also consider a roadside unit (RSU). A hybrid ap-

proach considers both, the coordination among the

CAVs and using a centralized system (typically RSU)

to create a global plan for the vehicles. The other

way of categorization is by implicit and explicit co-

ordination. In implicit coordination, the CAVs share

their intentions and desired maneuvers periodically

and have to deduce from the changed intentions of

the other CAVs whether their proposal has been ac-

cepted or not. Explicit coordination considers an ex-

plicit agreement with dedicated messages among the

vehicles to perform an acknowledged maneuver in an

event-based manner.

The present paper analyzes the state-of-the-art

for decentralized maneuver coordination that involves

only communication among the vehicles and does

not include the RSUs. Considering the existing

generic coordination approaches that are applicable

to a wide range of scenarios, we classify them into

four categories (Figure 1): Implicit Trajectory Broad-

cast (ITB), Explicit Trajectory Broadcast (ETB), Im-

plicit Trajectory Broadcast with Cost Values (ITB-

CV) and Space-Time Reservation (STR). For each

category, we present and analyze the respective publi-

cation. Then, we compare these approaches and ana-

lyze commonalities and differences. From the review

of the existing approaches, we derive future research

topics. We regard our systematic review as a contri-

bution for further research, standardization and devel-

opment of maneuver coordination for connected auto-

mated driving.

The remainder of the paper is structured as fol-

lows. Section 2 presents cooperative driving use cases

that beneﬁt from maneuver coordination. The cate-

gories and the selected approaches for decentralized

maneuver coordination are described in Section 3,

followed by their discussion and comparison in Sec-

tion 4, and a presentation of future research topics in

Section 5. Section 6 concludes the paper.

2 USE CASES FOR MANEUVER

COORDINATION

Cooperative driving brings many advantages for the

automated vehicles to achieve more comfortable,

safer and more efﬁcient driving, as well as to opti-

mize the trafﬁc ﬂow. It allows coordination of the

maneuvers in more speciﬁc use cases that can cause

conﬂicted and collision risk situations for the con-

ventional and automated vehicles. Several coopera-

tive use cases, where V2X communications can bring

beneﬁts and enable maneuver coordination, have been

A Survey on Decentralized Cooperative Maneuver Coordination for Connected and Automated Vehicles

101

Figure 2: Cooperative ACC.

Figure 3: Cooperative driving in a convoy.

identiﬁed e.g., by the R&D projects AutoNet2030 and

IMAGinE (Hobert et al., 2015; Llatser et al., 2019),

and are summarized as follows:

Cooperative-ACC (C-ACC). This use case extends

and improves the Adaptive Cruise Control (ACC) that

allows the vehicles to exchange additional informa-

tion using V2X to synchronize their velocities and

avoid more frequent acceleration and braking and in

the worst case, prevent critical situations (Figure 2).

C-ACC can also be used to enable platooning.

Cooperative Driving in a Formation. Vehicles driv-

ing in a platoon or convoy are considered as part of

formation driving. A platoon represents a group of

vehicles in a same lane on a highway, usually truck

platoons, that drive together in a stable formation

keeping small distances between each other, hence

increasing the road capacity, trafﬁc comfort and ef-

ﬁciency. A platoon has a master, typically the lead-

ing vehicle, that coordinates maneuvers with the other

vehicles and manages the platoon. Vehicles in a sin-

gle or multi-line convoy (Figure 3) is another way to

group the vehicles on highways. Instead of having

a group leader, the vehicle control is distributed over

all group vehicles in longitudinal and lateral direction

resulting in vehicle disturbances affecting all of the

convoy participants to a different extent; hence creat-

ing a stable formation. In a convoy, the vehicles only

need the neighboring vehicles’ dynamics information.

Cooperative Lane Change. In a cooperative lane

change situation, the vehicles share their planned ma-

neuver intentions and coordinate each other to suc-

cessfully change the lane (Figure 4). The cooperative

Figure 4: Cooperative lane change.

Figure 5: Cooperative driving at junction.

lane change maneuver can be executed between two

vehicles or within a group of few vehicles in a safe

and efﬁcient manner.

Cooperative Driving in Non-signalized

Intersections or Junctions. In intersections and

junctions without signalization, CAVs can coordinate

each other by exchanging their planned intentions

and safely execute turning maneuvers (Figure 5).

Cooperative Overtaking. Cooperative overtaking is

especially important on rural roads (Figure 6). Vehi-

cles can exchange and coordinate their future planned

trajectories to avoid a conﬂicted overtaking situation.

It can also be exploited by heavily loaded trucks on

highways to exchange their planned speed and current

weight for optimal coordination.

Cooperative Lane Merging. Lane merging is a com-

mon maneuver on highways. Lane merging also oc-

curs at construction site on the road. By exchanging

their intentions, vehicles can assist each other to coor-

dinate their merging maneuvers in a safe and efﬁcient

way (Figure 7).

Infrastructure-controlled Cooperative Driving.

The infrastructure can plan the trafﬁc distribution

to optimize the trafﬁc ﬂow and exchange the global

driving plan with the CAVs in different trafﬁc

situations. It can be used for trafﬁc intersection

management in both signalized and non-signalized

intersections or junctions to optimize the trafﬁc lights

and to manage the intersection passing of each CAV,

respectively (Figure 8). The infrastructure control

can also be used to coordinate the cooperative lane

merging.

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

102

Figure 6: Cooperative overtaking.

Figure 7: Cooperative lane merging.

3 DECENTRALIZED MANEUVER

COORDINATION

In this context, the decentralized maneuver coordina-

tion depends solely on the communication among the

CAVs to negotiate cooperative maneuvers. Further-

more, the coordination can be categorized into use

case-speciﬁc and generic. In a use case-speciﬁc co-

ordination, a coordination application is required that

focuses only on one speciﬁc trafﬁc situation, such as

the ones presented in Section 2, and uses protocol

only relevant to the respective use case. Generic co-

ordination aims at using one protocol to solve all co-

operative driving use cases.

3.1 Use Case-speciﬁc Coordination

Approaches

In order to solve the different trafﬁc situations among

the trafﬁc participants as described in Section 2,

a large number of research publications have con-

tributed to various coordination protocols for spe-

ciﬁc use cases. Platooning (Vukadinovic et al., 2018)

and C-ACC (Dey et al., 2016), which are based on

longitudinal coordination and utilize communications

among the vehicles, have been extensively investi-

gated and numerous publications on their character-

istics and control are available. Cooperative lane

changing and merging situations require lateral and

longitudinal coordination as vehicles also need to ac-

Figure 8: Cooperative intersection.

celerate or decelerate in order to create the needed

gap. Decentralized convoy driving allows the mem-

bers of a convoy to keep a stable formation by ad-

justing their longitudinal and lateral dynamics and

performing lane change maneuvers (Marjovi et al.,

2015). Cooperative lane change service outside of a

convoy as proposed in (Hobert et al., 2015) consists

of a search, preparation and execution phase. The ser-

vice supports maneuver negotiations and space reser-

vations using dedicated broadcast lane change mes-

sages, i.e., request, response, prepared or abort mes-

sages in each of the three phases. C-AAC can also

be used to achieve lane change and merging coor-

dination, see (Bevly et al., 2016) for an overview.

Distributed cooperative intersection and roundabouts

management without the need for infrastructure sup-

port is analyzed in (Chen and Englund, 2016), dis-

cussing distributed resource reservation protocols us-

ing different message sets. Instead of use case-

speciﬁc approaches, the review of the present paper

focuses on generic approaches which will be pre-

sented next.

3.2 Generic Coordination Approaches

The generic decentralized coordination represents

more recent state-of-the art approach considering ma-

neuver coordination protocols that are independent

of speciﬁc use case applications. They represent the

most promising concepts for maneuver coordination

and are presented in more detail in this work. Two

types of implicit and explicit coordination protocols

are discussed, as well as a hybrid approach incorpo-

rating infrastructure support for an already deﬁned

distributed protocol. The same lane merging sce-

nario is used to describe the different protocol pro-

posals consisting of four CAVs. The vehicles broad-

cast messages which consist of planned (PT), desired

(DT), alternative (AT) and requested (RT) trajecto-

ries, as well as space-time reservation (STR). It is im-

portant to mention that details of how the aforemen-

tioned trajectories are planned and generated are not

explained in the publications on trajectory broadcast

A Survey on Decentralized Cooperative Maneuver Coordination for Connected and Automated Vehicles

103

(a) Desired trajectory request

(b) Desired trajectory accepted

Figure 9: Maneuver coordination process in ITB.

approaches. In order to differentiate the unnamed dif-

ferent approaches, each of them is given a name and

the corresponding abbreviations are used to address

them (Figure 1).

3.2.1 Implicit Trajectory Broadcast (ITB)

The ﬁrst generic maneuver coordination protocol was

proposed by (Lehmann et al., 2018). It deﬁnes a new

message type, i.e., the Maneuver Coordination Mes-

sage (MCM) that carries trajectory information. In

the MCM, the data elements for trajectories are rep-

resented as Fren

et frames; a format that is commonly

used for trajectories along a geometric shape. The ve-

hicles periodically broadcast their planned trajectories

(PT) and optionally a desired trajectory (DT). The lat-

ter represents an alternative, more preferred trajectory

that is currently hindered by another vehicle due to the

right-of-way rules. In order to complete the maneuver

coordination, this work identiﬁes the following three

phases: detection, negotiation and execution.

Considering the lane merging scenario in Figure 9,

vehicle A detects the need for a lane merging maneu-

ver and shares its DT that currently intersects with the

PT of vehicle B (Figure 9a); in this way the negotia-

tion phase is started. After its own assessment, vehi-

cle B decides whether to accept or ignore the request.

In this situation, B accepts, adapts and broadcasts its

new PT. This enables the requesting vehicle A to exe-

cute its DT, which is now adapted into PT (Figure 9b).

In this implicit approach, the accepting vehicle ac-

knowledges that it accepts the requested DT by broad-

casting its new adapted PT, but does not explicitly re-

fer to a speciﬁc request. It is also required that each

(a) REQUEST and PROMISE

(b) CONFIRM

Figure 10: Maneuver coordination process in ETB.

vehicle that receives the DT broadcast has to deter-

mine whether its future planned maneuver intersects

this trajectory or not. In this situation, vehicles C and

D do not intersect the DT.

3.2.2 Explicit Trajectory Broadcast (ETB)

The idea of ITB was extended with a concrete explicit

coordination protocol that is also based on the princi-

ple of detection, negotiation and execution (Xu et al.,

2019). The emphasis is on the negotiation protocol,

which introduces a set of three message types: RE-

QUEST, PROMISE and CONFIRM. These messages

can be considered as MCM types and are broadcast

during the negotiation phase. They can carry multi-

ple trajectories in parallel, in contrast to ITB that con-

siders only one trajectory at a time. The protocol is

explained in the lane merging scenario in Figure 10

and the messages are also numbered to indicate their

order.

After vehicle A detected a need for cooperation, it

broadcasts a REQUEST message, which can consist

of multiple DTs. Multiple requests to different vehi-

cles are possible, too. In this situation, vehicles B and

C hinder this DT and after the DT request is received,

they decide whether to accept it or not. If the accept-

ing vehicles are able to plan collision-free trajectories,

they can send multiple alternative trajectories (AT) in

a PROMISE message (Figure 10a). The requesting

vehicle A constructs a collision-free global plan for all

of the participants and sends it via a CONFIRM mes-

sage. After that, in order to execute the desired ma-

neuver, the requesting vehicle needs one more broad-

cast message from the accepting vehicles that shows

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

104

that the vehicles adapted the promised PTs. If the ac-

cepting vehicles adapt their PTs as promised, the co-

ordination process is successful (Figure 10b), other-

wise after a certain timeout the requesting vehicle can

abort the coordination if one or all of the participating

vehicles are not able to adapt their promised PTs be-

cause of different reasons. The vehicles are also not

allowed to send different PROMISE and CONFIRM

messages within a certain period of time. In this way,

the protocol prevents potential ambiguities and risk of

divergence, which means that vehicles will not end up

choosing contradictory plans.

(Xu et al., 2019) also discuss communication fail-

ures, since the protocol does not assume a reliable

message transmission. In case of message loss, the

most important message is the CONFIRM message.

If any communication failure happens before, none of

the vehicles are going to change their trajectories. The

only problem can arise if the CONFIRM message is

delivered to a subset of vehicles. In the described sce-

nario, if the message is delivered to B, but not to C, B

will be the only vehicle to change the trajectory. This

can cause overhead but will not lead to a worse situa-

tion because the PROMISE and CONFIRM messages

have the requirement that only collision-free trajecto-

ries can be included.

3.2.3 Implicit Trajectory Broadcast with Cost

Values (ITB-CV)

(Llatser et al., 2019) propose an implicit coordina-

tion approach by periodically exchanging trajectories

with cost values. Similar to ITB, it relies on MCM as

the message type but assigns the cost values as addi-

tional trajectory attributes that express the necessity

and willingness of the CAVs for cooperation. The

proposed MCM format consists of three containers

describing the basic message information, the cur-

rent position of the vehicle and the trajectory infor-

mation. The cooperation protocol is explained in the

lane change scenario in Figure 11.

The protocol considers three different types of

trajectories: reference (PT), alternative (AT) and re-

quested (RT) trajectory. Each CAV sends its PT with

a cost value, representing its future planned trajectory.

In this situation, the PT of vehicle A has a cost value

C = 0.7. This means, A has a necessity to cooper-

ate since C > 0. The PT of the vehicle B has a cost

value of −0.2, indicating it is willing to cooperate

with other vehicles, because C < 0. If C = 0, the

vehicle does not have a necessity nor willingness to

cooperate. B sees the need of A and offers two ATs

with cost values 0.3 and 0.5 (Figure 11a). By deﬁ-

nition, these costs are higher than the reference cost

because in order to execute them, a coordination is

(a) Alternative trajectories offer

(b) Trajectories request

Figure 11: Maneuver coordination process in ITB-CV.

necessary. After A receives the ATs which represent

an offer from B, it sends two RTs with high willing-

ness costs: −1.0 and −0.8 (Figure 11b). The accept-

ing vehicle B constructs a global plan for both vehi-

cles by selecting its own trajectory and the trajectory

for the requesting vehicle A that gives the lowest to-

tal cost. Finally, the selected trajectories are adapted

as PTs and the cooperation is successfully completed

(Figure 11c).

3.2.4 Space-Time Reservation (STR)

The space-time reservation protocol (Heß et al., 2018;

Nichting et al., 2019; Heß et al., 2019) is a different

approach than ITB and ETB. It is not based on trajec-

tories exchange and request, but rather on an explicit

reservation of position and time constraints among the

communicating vehicles. This work merges a nom-

inal maneuver planner for autonomous driving with

a cooperative driving protocol. The space-time con-

straints are only exchanged once the need for cooper-

ation is identiﬁed using the containers of a CAM mes-

sage, hence using fewer messages with simple reser-

vation encoding.

A Survey on Decentralized Cooperative Maneuver Coordination for Connected and Automated Vehicles

105

(a) STR request

(b) STR accepted

Figure 12: Maneuver coordination process in STR.

In the lane change scenario in Figure 12, vehicle

A sends a REQUEST message that consists of the

STR constraints set with the following parameters:

position where the reservation should start, length

of the reserved area, time interval, velocity, ID of

the requesting vehicle and a reference to the corre-

sponding request (Figure 12a). If the accepting ve-

hicles can plan a collision-free maneuver incorporat-

ing these constraints, they send a COMMIT message

(Figure 12b). A REJECT message is sent if the vehi-

cles can not or does not want to accept the request. In

case the plans for the requesting vehicle changed, it

can also cancel the request. Figure 12 shows the tra-

jectories for better visibility; we note that they are not

periodically broadcast among the vehicles.

3.2.5 Infrastructure Support for Decentralized

Coordination

(Correa et al., 2019) present an enhancement of the

proposed ITB approach to consider infrastructure

support. Road side unit (RSU) information is in-

cluded in the MCM format alongside the proposed

trajectories exchange information between the vehi-

cles. In this way the infrastructure can support the

CAVs by offering advice on proposed speed, gap, lane

or transition of control between the automated system

and the driver using vehicle-to-infrastructure commu-

nications (Figure 8). This approach can lead to a more

neutral coordination, enhanced perception and coor-

dination involving multiple vehicles in speciﬁc trafﬁc

situations.

4 COMPARISON OF

APPROACHES

Table 1 shows a comparison of the presented decen-

tralized generic approaches using different criteria.

Since results from a performance evaluation are not

available for all of the described approaches or – if

available – the results are not comparable, the com-

parison is based on the conceptual design of the pro-

posed coordination protocols and on an analysis of

their advantages and shortcomings.

The ITB approach (Lehmann et al., 2018) brings a

lot of novelties, in particular the broadcast of planned

and desired trajectories, Maneuver Coordination Ser-

vice with Maneuver Coordination Messages focusing

only on the exchange of trajectories, as well as using

the Fren

et frames as standardized way to represent the

trajectories. However, considering more vehicles, this

protocol can result in ambiguities due to its implicit-

ness. In a situation with only two vehicles, the pro-

tocol can still cause conﬂicts due to the fact that it is

not known if the accepting vehicle changes its PT due

to the given request or because of another request. It

is also a serial coordination approach, meaning that at

a time the vehicles can only negotiate one trajectory.

This brings advantage in terms of reducing the motion

planning system complexity. However, a successful

coordination might require several requests, meaning

longer time to ﬁnd an acceptable DT. Considering all

these aspects, this concept stands as a solid basis for

cooperative maneuver coordination.

ETB (Xu et al., 2019) further improves the ETB

approach by proposing an explicit coordination with

a deﬁned number of negotiation messages that limits

communication failures and eliminates protocol am-

biguities. The approach also considers negotiation of

multiple trajectories and cooperation between more

than two vehicles. However, due to the higher number

of negotiation messages, the probability of communi-

cation failures is also increased.

The ITB-CV approach (Llatser et al., 2019) brings

an additional information to the trajectories by us-

ing cost values that help the requesting vehicles to

show the extent of their need for cooperation. It also

helps the accepting vehicle to decide whether to ac-

cept or reject a given request. However, the addition

of a cost value to each reference trajectory results in

a higher algorithmic complexity for the motion plan-

ning system, as it needs to be computed at each time

step. Since it is an implicit approach, the coordina-

tion process introduces the same ambiguities as in the

ITB approach. Also, compared to ETB, the impact

of communication failures will be larger because the

message losses can cause further ambiguities too.

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

106

Table 1: Comparison of generic approaches for decentralized maneuver coordination (BC = Broadcast, TR = Trajectory).

ITB ETB ITB-CV STR

(Lehmann et al., 2018) (Xu et al., 2019) (Llatser et al., 2019) (Heß et al., 2018)

Coordination type implicit explicit implicit explicit

Serial or parallel serial parallel parallel parallel

Number of vehicles 2 more than 2 2 more than 2

Communication type periodic BC periodic BC periodic BC non-periodic BC

Message type MCM MCM MCM CAM

Number of messages 2 4 3 2

Request method desired TR desired TR requested TR space-time

Protocol ambiguities yes no yes no

Impact of comm. failures yes limited yes limited

Simulation results no yes no yes

Experimental results no no no yes

A reservation of position and time constraints pro-

posed by the STR approach (Heß et al., 2018) offers

certain ﬂexibility as – in contrast to the other trajec-

tory request approaches – the requesting vehicle can

modify its planned trajectory within the reserved area.

The other main difference is the fact that the nego-

tiation is simpler because the accepting vehicles do

not broadcast their maneuver intentions. Once the ac-

cepting vehicles commit to the request, for the given

time they cannot intersect the requested constraints.

(Heß et al., 2018) is also the only publication that in-

tegrates the coordination process with an automated

motion planning system and validates the approach in

a real test environment using real test vehicles. Fur-

thermore, it is the only approach that uses an extended

CAM message only when the coordination need is de-

tected, hence requires the lowest bandwidth which is

another advantage in comparison with the other ap-

proaches that propose the periodic broadcast of a sep-

arate MCM message.

The ﬁrst three approaches utilize the broadcast of

trajectories, which improves the prediction system of

the CAVs. The best prediction of the other vehicles

movements is using their shared planned trajectory,

which reduces the uncertainty and improves the safety

in many difﬁcult or critical situations. By knowing

the intentions of the other vehicles, the CAVs will be

able to solve many situations. Disadvantages come

with the high data rate of periodic broadcasting and

the many implementation and communication issues

that need to be solved. The possibility to ﬁnd a faster

and more suitable solution increases with the parallel

approaches that allow negotiation of multiple trajec-

tories at a time between the vehicles; however the al-

gorithmic complexity of the motion planning system

also grows.

The implicit approaches might be an easier solu-

tion to implement in a simpler case with only two co-

operating vehicles because more difﬁcult situations or

more vehicles can bring various protocol and com-

munication failure ambiguities. In comparison, the

explicit approaches allow coordination between more

vehicles with limited impact of the communication

failures and no protocol ambiguities which makes it a

much safer solution that prevents conﬂicted situations

and additional risks. To enable fast, safe, efﬁcient and

unambiguous coordination for the cooperative vehi-

cles, the type and number of negotiation messages

play a crucial role. Table 1 shows the minimum re-

quired messages that allow the requesting vehicle to

execute the desired maneuver. This considers that the

coordination succeeds at ﬁrst try and all messages are

broadcast only once. The negotiation protocol pro-

posed in the ETB approach enables safe and unam-

biguous coordination with multiple options and more

than two included vehicles, but it also requires the

largest number of exchanged messages. The STR ap-

proach allows for fast explicit coordination with two

messages where the requesting vehicle does not need

to know the adapted trajectories of the other vehicles.

However, this might introduce ambiguity and some

overhead in the movement of the other participants,

especially in a more complex trafﬁc situation with

many vehicles included because it does not include a

ﬁnal conﬁrm message such as in ETB. A CONFIRM

message ensures that the selected trajectories of all

participating vehicles will not conﬂict.

Additional analysis is required to show the impact

of the presented approaches on the trafﬁc efﬁciency.

The ETB approach used the vehicular networking

simulation framework Artery

and evaluated the loss

of time caused by driving below the ideal speed in a

simple highway lane merging scenario. The results

have shown that the total time loss for the communi-

cating vehicles can be reduced up to 50% compared

https://github.com/riebl/artery

A Survey on Decentralized Cooperative Maneuver Coordination for Connected and Automated Vehicles

107

to non-communicating vehicles. The presented ap-

proaches also need to prove that they can prevent any

additional safety risks introduced by the maneuver co-

ordination process and validate the safety of the proto-

col using different metrics. (Correa et al., 2019) per-

formed a simulation in the microscopic trafﬁc simu-

lator SUMO

, which showed that this approach could

signiﬁcantly increase the safety by reducing the time-

to-collision (TTC) events with less than 3 s. In addi-

tion to the TTC, other safety metrics can be used such

as the Post Encroachment Time (PET) metric that de-

scribes how dangerous a certain situation can be.

All of the presented decentralized approaches in-

clude only a single requesting vehicle to initiate the

coordination process and the other vehicles need to

adapt based on their needs. The infrastructure sup-

port could help to provide more neutral coordination

in certain situations; a joint maneuver negotiation pro-

cess with more initiating vehicles can be considered

too. The more complex cascading process, where ve-

hicles need to send another maneuver request in order

to accept an incoming request, is avoided too.

5 OPEN TOPICS FOR

DECENTRALIZED MANEUVER

COORDINATION

The review and discussion in the previous sections

have shown that several approaches with different

characteristics exist. This section presents further re-

search gaps related to the detection and decision logic,

to the protocol and to V2X communications.

5.1 Detection and Decision Logic

How to Detect a Maneuver Coordination Need?

The reviewed approaches discuss only what happens

after the need for cooperation has been recognized

and the detection process is not described. An al-

gorithm is required that perceives the surrounding

CAVs hindering the desired maneuver and decides

when an alternative, more suitable and feasible ma-

neuver should be requested. Different metrics could

be used to take a decision such as improving the

time efﬁciency and avoiding safety-critical situation

in a worst-case scenario. Mixed trafﬁc scenarios

with communicating and non-communicating vehi-

cles should also be investigated, since the presented

approaches consider only situations involving CAVs.

How to Decide whether to Accept or Reject a Ma-

neuver Coordination Request? The best and easiest

https://www.eclipse.org/sumo

situation for a maneuver coordination is the one that

is beneﬁcial for all of the included vehicles. Since

in most of the situations one vehicle will be disad-

vantaged, the evaluation of the situation and request

is very important for the accepting vehicle and an

appropriate assessment is required. In (D

uring and

Pascheka, 2014), different types of cooperative and

uncooperative behavior are deﬁned based on a total

utility function. Metrics or cost functions consider-

ing loss of time, required deceleration and velocity or

potential safety critical consequences can be used to

decide whether to accept or reject a request in certain

trafﬁc situations.

5.2 Maneuver Coordination Protocol

Is an Application-independent, Robust

Representation of Trajectories Possible? Commu-

nicated trajectories need to be correctly interpreted

at both, the requesting and accepting vehicles. It

needs to be independent from speciﬁc applications

and a situation analysis system is needed to cor-

rectly represent the trajectories in the environmental

model of a CAV. Falsely interpreted or inaccurate

trajectory-related data can lead to conﬂicted negotia-

tion outcome for the involved vehicles and introduce

safety critical situations.

Can the Number of Involved Vehicles Be

Increased? A coordination between two vehi-

cles appears as a promising approach. Considering

the probability of successful cooperation and com-

munication failures, a coordination involving three

or more vehicles leads to a considerably higher

complexity and the protocols need to specify the

number of vehicles that could potentially cooperate

in more difﬁcult trafﬁc scenarios. The current

approaches have no upper bound on the potential

number of included vehicles and the scalability of

the coordination in different trafﬁc scenarios requires

further analysis.

What Kind of Message Type and Format has to

be Used? Most of the existing approaches propose

a new, dedicated message type (MCM) for the ex-

change of trajectory-related information among the

vehicles. RSU maneuver container in the MCM for-

mat is also discussed to incorporate the infrastruc-

ture support in speciﬁc situations, in this way utilizing

vehicle-to-infrastructure communications to enhance

the coordination process. The required standardized

format of trajectory representation will very much de-

pend on the data carried by the MCMs.

How Many Coordination Messages are Required?

Fast, safe, unambiguous and efﬁcient coordination re-

quires a certain ﬁxed number of negotiation messages

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

108

for each situation. Each coordination requires at least

a request and an acceptance or rejection message.

Final decision message such as the presented CON-

FIRM message (Xu et al., 2019) also ensures that the

coordination will be executed as planned. Additional

messages in speciﬁc situations might also be consid-

ered such as cancel message, if the requesting vehi-

cle decides to cancel the request, or abort message in

a situation when the requesting or accepting vehicle,

due to speciﬁc reasons, aborts an already agreed ma-

neuver in the execution phase.

Can Use Case-speciﬁc Application Messages Be In-

cluded? The proposed generic approaches cover sev-

eral cooperative maneuver coordination use cases but

some might need additional use case-speciﬁc infor-

mation. For this purpose, a generic protocol should

be able to incorporate application-speciﬁc messages.

Such type of messages could be required for the man-

agement of vehicles driving in a platoon or convoy,

or to request additional information required for the

completion of a speciﬁc maneuver such as the right

timing to perform a cooperative overtaking maneuver

on a rural road.

Which Message Generation Rules Can Be Iden-

tiﬁed? These rules deﬁne when and which vehicle

should send a message. They can have a huge im-

pact on the effectiveness of the coordination process

and data trafﬁc in general. Some of the reviewed ap-

proaches propose periodic broadcast, but the exact

interval is not deﬁned. Similar to CAM and CPM,

dynamic generation rules for maneuver coordination

messages could be considered where the message in-

terval depends on the vehicle dynamics, i.e., speed,

heading and acceleration.

How Can Cascading Be Enabled? Maneuver cas-

cading has so far been avoided by the presented de-

centralized approaches. It can help in many trafﬁc

situations to realize a requested maneuver. The ac-

cepting vehicle needs to request a maneuver itself to

another adjacent vehicle. If this additional maneuver

is successful, assuming that the current driving situ-

ation did not change signiﬁcantly, it will enable the

initial requesting vehicle to execute its desired maneu-

ver. It can also be seen as an explicit maneuver coor-

dination between more than two vehicles, which will

include additional negotiation messages. Such a cas-

cading maneuver will prolong the negotiation process

and will bring additional complexity, but it can even-

tually increase the probability of a successful coordi-

nation. A further analysis is needed to show whether

such a maneuver can be safe and efﬁcient enough to

be considered as an addition to the coordination pro-

tocol in speciﬁc situations.

Can the Data Security and Privacy Be Guaran-

teed? It can be presumed that maneuver coordina-

tion will apply digital signatures and certiﬁcates of the

V2X communication system that provides integrity,

authentication and non-repudiation of the exchanged

messages. Still, open challenges exist, e.g. for mis-

behavior detection and mitigation as an application-

speciﬁc security mechanism. Similarly, privacy will

be expected to rely on short-living and changing

pseudonyms. However, pseudonyms must not be

changed during a maneuver since it is a safety-critical

situation. Also, the small number of vehicles involved

in a maneuver may undermine the anonymity since

the requesting or accepting vehicle may be identiﬁ-

able.

5.3 V2X Communications

What are the Communication Requirements for

Maneuver Coordination? So far, the V2X com-

munication system has been primarily designed for

driver information and warnings with relaxed com-

munication requirements. It is commonly accepted

that safety-critical communications such as maneuver

coordination require very low latency (∼ 10 ms) and

very high communication reliability (> 99%) (Boban

et al., 2018). The speciﬁc requirements for the ex-

change of multiple subsequent messages are not yet

well understood since most of the existing work refers

to individually broadcast messages.

Can Advanced Features of the Underlying Ac-

cess Technology Be Exploited? WLAN-V2X and

Cellular-V2X have been widely studied and their po-

tential beneﬁt for safety applications is well investi-

gated. It is still to be seen whether the speciﬁc ad-

vanced features in Cellular V2X translate it into im-

proved performance for maneuver coordination. One

example of these features is the bounded latency of

Sensing-based Semi-Persistent Scheduling (SB-SPS)

in Cellular-C2X in scenarios with a high network

load. Also, it is to be investigated whether the evolu-

tion of the access technologies, incl. IEEE 802.11bd

and 5G NR V2X bring advantages, e.g. for the relia-

bility or the latency of the message exchange.

Will Broadcast Communication Prevail? V2X

communication is primarily based on broadcast com-

munication, more speciﬁcally single-hop broadcast

or (in the European V2X system) multi-hop broad-

cast within a deﬁned geographical area. By de-

sign, broadcast does not provide reliable message

exchange since the feedback implosion prevents ap-

plying acknowledgements and re-transmissions. The

message exchange for cooperative maneuver coordi-

nation typically involves only few vehicles and may

A Survey on Decentralized Cooperative Maneuver Coordination for Connected and Automated Vehicles

109

facilitate other approaches than broadcast, e.g., small-

group multicast with explicit acknowledgment that in-

creases the reliability.

Should Multi-channel Operation Be Applied? A

higher number of messages increases the channel

load, which results in a lower reliability and longer

latency. In order to reduce the risk of channel con-

gestion, an analysis is required whether the MCMs

should be integrated on the same channel with other

messages with a prioritization among the messages,

or a multi-channel option should be considered.

6 CONCLUSION

Maneuver coordination using V2X communications

targets at safer, more comfortable and efﬁcient driv-

ing for CAVs. Generic approaches for maneuver

coordination can be identiﬁed as a research trend.

These approaches solve different trafﬁc situations by

a scenario-independent solution. In the present paper,

existing proposals were reviewed and analyzed. Also,

seven use cases, for which maneuver coordination is

expected to bring beneﬁts, were presented, ranging

from C-ACC to infrastructure-controlled cooperative

driving. In order to explain the differences and novel-

ties, the existing generic approaches for decentralized

coordination were described in detail for a lane merg-

ing scenario as an example use case. The approaches

were classiﬁed into four categories: Implicit Trajec-

tory Broadcast (ITB), Explicit Trajectory Broadcast

(ETB), ITB with Cost Values (ITB-CV) and Space-

Time Reservation (STR).

The analysis and discussion of the proposed pro-

tocols in the paper has shown that explicit maneuver

negotiation and broadcast of future maneuver inten-

tions can enable safe and efﬁcient maneuver coordi-

nation. Further analysis is needed to evaluate the im-

pact of the proposed approaches on trafﬁc safety and

efﬁciency. The introduction of additional safety risks

needs to be eliminated. Well-deﬁned metrics for traf-

ﬁc safety and efﬁciency should be considered to as-

sess the performance. In the paper, challenges were

highlighted and future research directions identiﬁed.

These include the detection and decision logic of ma-

neuvers, syntax and semantic of the maneuver pro-

tocol as well as reliability mechanisms of the V2X

protocol. The key research question is: How can a

use case-independent, reliable and low-latency proto-

col for safe, comfortable and efﬁcient maneuver coor-

dination be designed?

ACKNOWLEDGEMENTS

This work was gratefully supported by the German

Science Foundation (DFG) by project KOALA 2

under number 273374642 within the priority pro-

gram Cooperatively Interacting Automobiles (CoIn-

Car, SPP 1835). The authors thank Daniel Heß (DLR)

and Alexey Vinel (Halmstad University) for valuable

discussions. The scenario ﬁgures in the paper were

created with the illustration toolkit from the C2C-

CC.

REFERENCES

Bevly, D. et al. (2016). Lane change and merge maneu-

vers for connected and automated vehicles: A survey.

IEEE Transactions on Intelligent Vehicles, 1(1):105–

120.

Boban, M., Kousaridas, A., Manolakis, K., Eichinger, J.,

and Xu, W. (2018). Connected roads of the future:

Use cases, requirements, and design considerations

for vehicle-to-everything communications. IEEE Ve-

hicular Technology Magazine, 13(3):110–123.

Chen, L. and Englund, C. (2016). Cooperative intersection

management: A survey. IEEE Transactions on Intel-

ligent Transportation Systems, 17(2):570–586.

Correa, A. et al. (2019). Infrastructure support for cooper-

ative maneuvers in connected and automated driving.

In 2019 IEEE IV Symposium, pages 20–25.

Dey, K. C. et al. (2016). A review of communication, driver

characteristics, and controls aspects of Cooperative

Adaptive Cruise Control (CACC). IEEE Transactions

on Intelligent Transportation Systems, 17(2):491–509.

uring, M. and Pascheka, P. (2014). Cooperative decen-

tralized decision making for conﬂict resolution among

autonomous agents. In 2014 IEEE INISTA, pages

154–161.

Heß, D., Lattarulo, R., P

erez, J., Hesse, T., and K

oster, F.

(2019). Negotiation of cooperative maneuvers for au-

tomated vehicles: Experimental results. In 2019 IEEE

ITSC, pages 1545–1551.

Heß, D., Lattarulo, R., P

erez, J., Schindler, J., Hesse, T.,

and K

oster, F. (2018). Fast maneuver planning for

cooperative automated vehicles. In 2018 ITSC, pages

1625–1632.

Hobert, L. et al. (2015). Enhancements of V2X communi-

cation in support of cooperative autonomous driving.

IEEE Communications Magazine, 53(12):64–70.

Lehmann, B., G

unther, H., and Wolf, L. (2018). A generic

approach towards maneuver coordination for auto-

mated vehicles. In 2018 ITSC, pages 3333–3339.

Llatser, I., Michalke, T., Dolgov, M., Wildsch

utte, F., and

Fuchs, H. (2019). Cooperative automated driving use

cases for 5G V2X communication. In 2019 IEEE 2nd

5G World Forum, pages 120–125.

URL: http://https://www.car-2-car.org

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

110

Marjovi, A., Vasic, M., Lemaitre, J., and Martinoli, A.

(2015). Distributed graph-based convoy control for

networked intelligent vehicles. In 2015 IEEE IV Sym-

posium, pages 138–143.

Molina-Masegosa, R. and 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 Maga-

zine, 12(4):30–39.

Nichting, M., Heß, D., Schindler, J., Hesse, T., and K

oster,

F. (2019). Explicit negotiation method for cooperative

automated vehicles. In 2019 ICVES, pages 1–7.

oberg, K., Andres, P., Buburuzan, T., and Brakemeier, A.

(2017). Cooperative Intelligent Transport Systems in

Europe: Current deployment status and outlook. IEEE

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

Vukadinovic, V. et al. (2018). 3GPP C-V2X and

IEEE 802.11p for vehicle-to-vehicle communications

in highway platooning scenarios. Elsevir Ad Hoc Net-

works, 74:17 – 29.

Xu, W., Willecke, A., Wegner, M., Wolf, L., and Kapitza,

R. (2019). Autonomous maneuver coordination via

vehicular communication. In 2019 IEEE/IFIP DSN-

W, pages 70–77.

A Survey on Decentralized Cooperative Maneuver Coordination for Connected and Automated Vehicles

111