《PIARC :自动驾驶车辆-道路运营商和道路管理部门面临的挑战和机遇(英文版)(88页).pdf》由会员分享,可在线阅读,更多相关《PIARC :自动驾驶车辆-道路运营商和道路管理部门面临的挑战和机遇(英文版)(88页).pdf(88页珍藏版)》请在三个皮匠报告上搜索。
1、 www.piarc.org 2021R03EN AUTOMATED VEHICLES CHALLENGES AND OPPORTUNITIES FOR ROAD OPERATORS AND ROAD AUTHORITIES TASK FORCE B.2 AUTOMATED VEHICLES:CHALLENGES AND OPPORTUNITIES FOR ROAD OPERATORS AND ROAD AUTHORITIES 2021R03EN AUTOMATED VEHICLES STATEMENTS The World Road Association(PIARC)is a non-pr
2、ofit organisation established in 1909 to improve international co-operation and to foster progress in the field of roads and road transport.The study that is the subject of this report was defined in the PIARC Strategic Plan 2016 2019and approved by the Council of the World Road Association,whose me
3、mbers are representatives of the member national governments.The members of the Technical Committee responsible for this report were nominated by the member national governments for their special competences.Any opinions,findings,conclusions and recommendations expressed in this publication are thos
4、e of the authors and do not necessarily reflect the views of their parent organisations or agencies.This report is available from the internet site of the World Road Association(PIARC):http:/www.piarc.org Copyright by the World Road Association.All rights reserved.World Road Association(PIARC)Arche
5、Sud 5 niveau 92055 La Dfense CEDEX,FRANCE International Standard Book Number:078-2-840606-653-6 Front cover AdobeStock AUTOMATED VEHICLES CHALLENGES AND OPPORTUNITIES FOR ROAD OPERATORS AND ROAD AUTHORITIES TASK FORCE B.2 AUTOMATED VEHICLES-CHALLENGES AND OPPORTUNITIES FOR ROAD OPERATORS AND ROAD AU
6、THORITIES 2021R03EN AUTOMATED VEHICLES AUTHORS/ACKNOWLEDGEMENTS This report has been prepared by the working group Task Force“Automated vehicles:challenges and opportunities for road operators and road authorities”of the Task Force B.2 of the World Road Association(PIARC).The contributors(alphabetic
7、al order)to the preparation of this report are:Martin Bhm(Austria)Rudi Botha(South Africa)Domenico Crocco(Italy)Luigi Carrarini(Italy)Ugo Dibennardo(Italy)Jian Gao(China)Wayne George(United Kingdom)Matthew Hall(Australia)Darina Havlickova(Czech Republic)Abdelmename Hedhli(France)Yuji Ikeda(Japan)Ana
8、 Luz Jimnez Ortega(Spain)Jiwoon Kang(South Korea)Matt Krech(Canada)Eric Ollinger(France)Ian Patey(United Kingdom)Hirotaka Sekiya(Japan)Martin Thibault(Canada-Qubec)Cen Yanqing(China)Petr Zamecnik(Czech Republic)Hamid Zarghampour(Sweden)The Task Force B.2 was chaired by Eric Ollinger(France)and Abdel
9、mename Hedhli(France),Ian Patey(United Kingdom)and Ana Luz Jimnez Ortega(Spain)were respectively the French,English and Spanish-speaking secretaries.EXECUTIVE SUMMARY 2021R03EN AUTOMATED VEHICLES CHALLENGES AND OPPORTUNITIES FOR ROAD OPERATORS AND ROAD AUTHORITIES Vehicle Automation at SAE levels 3,
10、4 and 5 is still in a testing phase in most countries.Many tests at level 3 have been performed successfully on open roads without any changes to the road infrastructure and without any observed impacts on congestion.There are various situations which necessitate that a driver takes control of the v
11、ehicle a fall-back solution that is still allowed at level 3,but not at level 4 or 5.It is likely that some support from the infrastructure will be needed to reach higher levels of automation.In particular,well-maintained lane markings and the provision of landmarks appear to be key features.Automat
12、ed vehicles cannot,however,rely solely on the physical infrastructure,due to practical limitations;lane markings deteriorate over time,and it is not practical to forecast when they become ineffective for automated operation.Therefore,digital infrastructure,including data provided through high-defini
13、tion maps and/or through vehicle-to-infrastructure connectivity,is required.Initial insights on this digital infrastructure have been provided in this report.Regarding the impact on congestion,two factors need to be considered.The first is the operational capacity of physical infrastructure.Some stu
14、dies indicate that capacity is likely to decrease at low penetration rates and potentially improve only by 2050 or 2055 due to shorter headways.The second is the change in traffic demand.It has been predicted that within the next 20 years,60%of the the worlds population will live in cities.If vehicl
15、e use is increased without regulation due to increasing empty automated vehicle trips and changing parking availability,congestion could increase further.The promotion of automated shuttles could help address this risk.Some studies also show that the impact on road safety should be positive at high
16、penetration rates,as already observed due to greater penetration of Advanced Driver Assistance Systems(ADAS).This assumes that many human factors are considered,such as driver overreliance or loss of driving skills.The report mentions 11 key challenges to tackle regarding social issues in order to m
17、ake automation a success.Finally,in terms of responsibility,a new balance will need to emerge between the responsibility of the road operator and the responsibility of car manufacturers(or even the car itself).This balance could be linked to a certain level of service,but the definition of this leve
18、l is not yet mature.AUTOMATED VEHICLES 2021R03EN 2 CONTENT 1 INTRODUCTION.6 2 METHODOLOGY.8 3 PHYSICAL INFRASTRUCTURE.9 3.1 ROAD SIGNS.9 3.2 STATIC AND DYNAMIC SIGNS.10 3.3 LANE MARKINGS:NOT A ROBUST SYSTEM VS GNSS POSITIONING.13 3.4 NEED FOR LANDMARKS.14 3.5 ROAD GEOMETRY,INFRASTRUCTURE DESIGN.14 3
19、.6 PLATOONING:IMPACT ON THE INFRASTRUCTURE.19 3.7 CITY PLANNING.20 3.8 MAINTENANCE STRATEGIES.21 4 DIGITAL INFRASTRUCTURE-CONNECTIVITY.25 4.1 NECESSITY OF CONNECTIVITY.25 4.2 USE CASES REQUIRING CONNECTIVITY.26 4.3 REQUIREMENTS FOR CONNECTIVITY.33 4.4 COMMUNICATION MEASURES.34 5 DIGITAL INFRASTRUCTU
20、RE DIGITAL MAPS AND POSITIONING.36 5.1 INTRODUCTION.36 5.2 DIGITAL MAP STANDARDS.38 5.3 DIGITAL TWIN.38 5.4 POSITIONING.39 5.5 SATELLITE BLIND SPOTS.41 6 DIGITAL INFRASTRUCTURE DATA ISSUES COMMON TO CONNECTIVITY AND DIGITAL MAPS.42 6.1 DATA MANAGEMENT FOR ROAD NETWORK OPERATORS.44 6.2 DATA ACCESS.46
21、 7 IMPACTS ON ROAD NETWORK OPERATIONS(RNO).52 7.1 IMPACT OF AUTOMATION ON ROAD NETWORK OPERATIONS.52 AUTOMATED VEHICLES 2021R03EN 3 7.2 IMPACTS ON TRAFFIC OPERATIONS AND ROAD CAPACITY.52 7.3 PLATOONING.56 7.4 AUTOMATED SHUTTLE MUST BE CONSIDERED IN AN URBAN AND RURAL ENVIRONMENT.58 7.5 HOW RNO CAN H
22、ELP CAV ON WORK ZONES/INCIDENTS.60 7.6 TRAFFIC MANAGEMENT BY RNO.60 8 RESPONSIBILITY AND FINANCING.62 8.1 RESPONSIBILITY AND INSURANCE.62 8.2 COSTS,BENEFITS AND FINANCING.63 9 SOCIAL ISSUES AND THE CONSIDERATION OF HUMAN FACTORS.65 9.1 MAIN PRINCIPAL QUESTIONS FROM THE PERSPECTIVE OF SOCIETY.65 9.2
23、THE LIMITATION OF POTENTIAL BENEFITS OF AV BY A RANGE OF HUMAN FACTORS.66 9.3 NEW HUMAN FACTORS CAN ARISE.67 9.4 RECOMMENDATIONS FROM THE PERSPECTIVE OF SOCIETY.71 10 CONCLUSION.73 11 RECOMMENDATIONS.74 11.1 RECOMMENDATIONS FOR DECISIONS MAKERS.74 11.2 RECOMMENDATIONS FOR ROAD OPERATORS AND ROAD AUT
24、HORITIES.76 11.3 RECOMMENDATIONS FOR LMICS.76 11.4 RECOMMENDATIONS FOR PIARC.77 12 GLOSSARY.78 13 REFERENCES.80 AUTOMATED VEHICLES 2021R03EN 4 LIST OF FIGURES Figure 1:SAE J3016 Levels of Driving Automation 33.6 Figure 2:Examples of variances in Vienna Convention implementations 2.9 Figure 3:Example
25、s of variances in MUTCD-influenced implementations.10 Figure 4:Variable message sign(VMS)source:MTO.11 Figure 5:Changeable message sign(CMS).11 Figure 6:Variable Speed Limit sign(VSLS).11 Figure 7:Lane Use Management Sign(LUMS).11 Figure 8:Examples of the effect of LED refresh rates-blackened sectio
26、ns.12 Figure 9:Magnetic markers and magnetic-induction lines(Source:NILIM,MLIT,Japan).14 Figure 10:Example of automated low-speed shuttle landmark sign.14 Figure 11:Landmark positioning as used in Germany.14 Figure 12:Pavement marking.15 Figure 13:Example of continuous hard shoulder(left)and Emergen
27、cy Refuge Area(right).17 Figure 14:LIDAR rendering of a snowfall-numerous of purple cloud point(source:Waymo).19 Figure 15:Bushes visible on street side(Source:NILIM,MLIT,Japan).22 Figure 16:Example of foliage blocking a low-speed shuttle landmark 4 22 Figure 17:Road width in snowy conditions(Source
28、:NILIM,MLIT,Japan).23 Figure 18:Examples of pavement marking and road sign maintenance issues 14.24 Figure 19:Limitation of range by sensor detection.25 Figure 20:Information on-board sensors are not able to detect.25 Figure 21:Data fusion 16.26 Figure 22:Vehicle to Everything(V2X)17.26 Figure 23:Ro
29、ad obstacle information provision service.27 Figure 24:Work zone USDOT Project phases(WZDx).28 Figure 25:Congestion information provision service.28 Figure 26:Tollgate information provision service.29 Figure 27:Merging support service.29 Figure 28:Left Turn Assist(LTA).30 Figure 29:Vehicle Turning R
30、ight in Front of Bus Warning.31 Figure 30:Red Light Violation Warning(RLVW).31 Figure 31:Spot Weather Impact Warning(SWIW).32 AUTOMATED VEHICLES 2021R03EN 5 Figure 32:Queue Warning(Q-WARN).32 Figure 33:ITS information services at an intersection 20.33 Figure 34:Concept of merging support service.34
31、Figure 35:Automated Driving 22.35 Figure 36:Platooning 22.35 Figure 37:Remote Driving 22.35 Figure 38 Level model of a Local Dynamic Map 25.37 Figure 39:In-vehicle process to create an environmental model.42 Figure 40:Levels of the Infrastructure Support for Automated Driving 33.43 Figure 41:Example
32、s of ISAD levels along the road network 33.44 Figure 42:Stakeholders in RNO data exchange 36.46 Figure 43:Simulation results representing the percentage change in capacity of each freeway segment relative to the base scenario.54 Figure 44:Impacts on AVs on the freeway capacity in Germany.55 Figure 4
33、5:Automated driving of shuttle in rural area(Kamikoani village,Japan).59 AUTOMATED VEHICLES 2021R03EN 6 1 INTRODUCTION In the last few years,vehicles with different levels of automation have been being tested all over the world.The Society of American Engineers has identified five levels of driving
34、automation(SAE levels):Figure 1:SAE J3016 Levels of Driving Automation 33 SAE levels 1 and 2 are driver support features,many have been authorized and deployed in a number of countries.Levels 3 to 5 represent automated driving features and are still being tested by car manufacturers with prototypes
35、using different combinations of sensors,Global Navigation Satellite System(GNSS)positioning and sometimes connectivity.An important aspect to consider is the Operational Design Domains(ODD)for which a feature has been designed to operate within.ODDs are defined in the same SAE standard as“operating
36、conditions under which a given driving automation system or features thereof are specifically designed to function,including but not limited to,environmental,geographical,and time-of-day-restrictions,and/or the requisite presence or absence of certain traffic or roadway characteristics.”This include
37、s roadway types,speed range,environmental conditions(weather,daytime/night-time,etc.),and prevailing traffic law and regulations.Road operators and road authorities are keen to understand the impacts of automated vehicles(at different SAE levels)on traffic management,equipment requirements,road safe
38、ty,maintenance strategies,adaptive control and performance optimization of the network and infrastructure design.This report provides the current state of understanding for all these questions,based on the experience gathered from testing and experience with vehicle automation all around the world.A
39、UTOMATED VEHICLES 2021R03EN 7 The report provides an initial explanation of the methodology used for the gathering and synthesis of the content.The implications of existing physical infrastructure and its condition on the effectiveness of automation is discussed as well as the impact of automation o
40、n the physical infrastructure.Irrespective of the quality and condition of physical infrastructure,it is evident that automated vehicles will increasingly need to rely on a digital twin of this infrastructure,comprising data transmitted either through connectivity or through digital maps.This is dis
41、cussed in the chapters concerned with developing this necessary digital infrastructure.Also discussed is the impact of automation on road network operations and traffic management,with implications for road network planning and road agency capabilities.The closing chapters tackle legal,economic(resp
42、onsibility,insurance and financing)and social issues(human factors,including road safety).AUTOMATED VEHICLES 2021R03EN 8 2 METHODOLOGY In producing this report,the Task Force firstly defined the main challenges and opportunities that road network operators and road authorities are likely to face fro
43、m automated vehicles.An initial workshop meeting defined the main topics for the Task Force to consider these topics form the chapters of this report.The report begins with a consideration of the implications and responsibilities for road network operators and authorities on both physical and digita
44、l infrastructure.The following aspects of physical infrastructure have been considered:road signs,lane markings,static and dynamic signs and road geometry,and city planning and maintenance;including the potential changes to infrastructure design to facilitate platooning.Digital infrastructure was id
45、entified as a significant area of interest,with consideration given to digital maps,positioning aspects and data.It was recognised that automated vehicles will impact non-technical areas such as responsibilities and insurance as well as social issues.The report concludes with several recommendations
46、 including some specific to low-and middle-income countries.An extensive literature search was undertaken to understand automated vehicle activity projects,research,trials,and policy development across the globe.This was followed by workshop sessions which identified a list of existing case studies
47、to be utilised for source material-from Australia,Austria,Canada,China,Czech Republic,France,Italy,Japan,Kenya,Korea,Spain,South Africa,Sweden,and the United Kingdom.A cross fertilisation session was also held with platooning experts from Europe to enrich the report during the PIARC World Congress h
48、eld in Abu Dhabi.A core team of writers was assigned to each chapter,with reviews by the rest of the Task Force members-to discuss and enrich the content and perspective.This was followed by a consolidation and editing process to finalize the report.AUTOMATED VEHICLES 2021R03EN 9 3 PHYSICAL INFRASTR
49、UCTURE 3.1 ROAD SIGNS 3.1.1 Harmonization of road signs Road signs are a critical aspect of a safe road network,aligned with the ability of drivers to recognise and understand them.Human drivers are relatively adaptable in their ability recognise and understand signs that look familiar even if not e
50、xactly as they had expected.Machines are not necessarily as adaptable.Acknowledging that vehicles are manufactured for global markets,significant international efforts are needed to harmonize traffic signs and road markings to ensure consistent machine recognition and safety 1 performance.Due to eco
51、nomies of scale,the automotive manufacturing industry has limited ability to customize software and sensor system design for each individual jurisdiction.Therefore,industry-government forums are needed at an international level to support traffic sign harmonization.The United Nations Economic Commis
52、sion for Europe has a working Group on harmonization 2.There are two major standards for traffic sign harmonization:the US Manual on Uniform Traffic Control Devices(MUTCD)for Streets and Highways and the Vienna Convention on Road Signs and Signals.There is also the Southern African Development Commu
53、nity Convention and the Central American Integration System Convention.Many countries also use individual systems that combine design principles from both the MUTCD and Vienna Convention.It is also common to have regional differences in signage within countries and even within sub-jurisdictions as w
54、ell as on private land.Many road authorities have national standards for road signs,but regulation and enforcement are left to local governments.Road authorities that have a national agency responsible for implementing and approving signage have demonstrated significantly more uniform systems 3.Figu
55、re 2:Examples of variances in Vienna Convention implementations 2 Some of the key differences and challenges for harmonization include:Units-use of the metric system vs.imperial system;some countries do not indicate units on road signs Shape use of diamond,pentagonal or triangular signs for various
56、warning purposes Colour use of orange for temporary signage in MUTCD,use of blue for mandatory information in Vienna Convention vs.supplementary information in MUTCD Language use of text in various languages AUTOMATED VEHICLES 2021R03EN 10 Figure 3:Examples of variances in MUTCD-influenced implement
57、ations The benefits from investing in road sign harmonization include improved machine vision system and algorithm performance as well as limiting barriers to higher levels of ODD implementation and testing.There are also the benefits of reduced need for costly alternatives(digitization),improved hu
58、man recognition and opportunities to improve compliance through more consistent and reliable in-vehicle alerts.It is recommended that priority is given to resolving traffic sign issues on expressways/motorways first as this is where the greatest volume of traffic occurs and where higher levels of au
59、tomated driving are most likely to appear first 3.Harmonization efforts should also focus on sign shape which is the primary method of recognition as Traffic Sign Recognition(TSR)systems do not distinguish colour well 3.Harmonization will take time,and even after a consensus has been reached,physica
60、lly changing signs along every road worldwide will be a significant challenge.Meanwhile,vehicle manufacturers and original equipment manufacturers(OEMs)are introducing different national sign databases in their systems.3.1.2 Alternative approaches adding digitization/connectivity The Task Force note
61、d that some jurisdictions are taking an alternative approach to prepare for higher levels of automation.Rather than waiting for harmonization efforts to advance,jurisdictions may select to digitize all forms of signage by adding infrastructure-to-vehicle(I2V)connectivity or machine-readable code(e.g
62、.,Quick Response(QR)code).The HMI(Human Machine Interface)needs to be properly defined.With millions of road signs in each jurisdiction,harmonization should be viewed as a long-term goal.Notwithstanding,there are additional benefits in redundancy from having both harmonization and digitization of ro
63、ad signage.3.2 STATIC AND DYNAMIC SIGNS As noted in the previous section,there are multiple sign standards around the world for static signs.Most are inspired by either the Vienna convention from 1968 or the US MUTCD first published in 1935.AUTOMATED VEHICLES 2021R03EN 11 AV sign recognition systems
64、 need to read all these signs,whether they are static,dynamic or electronic signs.There are four main types of dynamic signs currently in use on roads and highways,built by many manufacturers to a variety of standards and specifications.Figure 4:Variable message sign(VMS)source:MTO Figure 5:Changeab
65、le message sign(CMS)Figure 6:Variable Speed Limit sign(VSLS)Figure 7:Lane Use Management Sign(LUMS)3.2.1 Traffic Sign Recognition Traffic Sign Recognition(TSR)is a technology that enables a vehicle to detect and recognize traffic signs at the roadside e.g.speed limit or children or turn ahead.This i
66、s an example of an Advanced Driver Assistance System(ADAS).The technology is being developed by a variety of automotive suppliers.Vehicle manufacturers are moving towards enabling speed assistance systems and automated driving systems which use TSR technology;the benefits of successful introduction
67、are likely to be significant for road safety.TSR uses image processing techniques to detect and recognize traffic signs.The detection methods can be generally divided into colour based,shape-based and learning based methods.An AV equipped with a TSR system can read and recognize traffic signs.Reliab
68、ility for the reading and recognition of standard static signs(speed limit,warnings,etc.)by TSR is very high;in Australia and in France pilot projects have demonstrated a recognition rate of these signs of almost 100%.However,TSR systems cannot yet read and recognize way finding signs,non-standard a
69、nd information/advisory signs.AUTOMATED VEHICLES 2021R03EN 12 TSR systems rely on signs to be correctly located and maintained(including cleaning)so that visible light and colour can be captured by cameras.3.2.2 Traffic Sign Recognition system with dynamic signs The case studies collated for this re
70、port indicate that TSR systems are currently unable to consistently read dynamic signs(roadside signs or gantry mounted sign).Literature and stakeholder interviews indicated that the refresh rate of signs and variability of pixel illumination could vary between brands and designs.Other factors could
71、 include the sign size,height and approach angle,as well as level of illumination from the power source.3.2.3 CMS,VMS,VSLS and LUMS signs TSR do not generally recognize Variable message signs(VMS)and Changeable Message Sign(CMS)systems as they tend to focus on fixed speed sign recognition.Variable S
72、peed Limit Signs(VSLS)and Lane Use Management Signs(LUMS)also are not recognized by current TSR systems.3.2.4 Refresh rate The refresh rate of electronic signs is designed to enable a human eye to see them properly,without flickering.It is necessary to develop a standard that enables a TSR system to
73、 do the same.Different standards currently exist,with both the New Zealand Standard and EU guidelines recommend that the frequency of emitted light should be“not less than 90 Hz”while the Australian standard is a significantly higher 2 kHz.Flickering in the electronic display may be observed by the
74、TSR systems camera causing it difficulties in recognizing the sign.This occurs on some electronic signs and not others.In some cases,segments of the sign may have refresh rates out of sequence with other segments of the sign.Evidence suggests this flickering effect may not be apparent in direct curr
75、ent powered signs,or signs from certain sign manufacturers.This could be improved either by improving either the signs or the cameras.As cameras are getting better year by year the recommended priority would be with camera improvements.3.2.5 IVS(In Vehicle Signage)Figure 8:Examples of the effect of
76、LED refresh rates-blackened sections.AUTOMATED VEHICLES 2021R03EN 13 A possible solution to the difficulty of detection of fixed or variable signs by TSR systems is to use infrastructure-to-vehicle(V2I)connectivity.With the help of RSUs(Road Side Units)and OBUs(On Board Units)in vehicles,the signs c
77、an be identified,read and understood by equipped vehicles.This solution added to TSR systems can improve sign detection,interpretation and response.3.3 LANE MARKINGS:NOT A ROBUST SYSTEM VS GNSS POSITIONING 3.3.1 Role and challenges of lane markings Automated vehicle function currently relies on clea
78、r,consistent and well-maintained road markings and signs to navigate.Lane markings are one of the main reference elements for automated vehicles in maintaining their position on the road and within lanes.However,current lane markings do not fully meet the needs of automated vehicles.For example,ther
79、e are various road layouts and situations(road works,tolling plazasetc.)where there are no lane markings.The regulations for the secondary road network in Austria do not require lane markings to be provided.Weather can cause issues,as some lane markings can be obscured by rain or when the sun is at
80、low angles.To support current levels of automation,the assessment and maintenance of road markings(including testing and assurance of retro-reflectivity)is required.One of the main challenges for road operators is that it is technically impossible to predict or detect the“failure”of markings(degrada
81、tion to a point that automation is compromised).It is inevitable that automated vehicles will have to cope damaged or absent markings,which may need supplementary information through digital maps.These digital maps will also be required to be maintained,either manually or through a semi-automated pr
82、ocess.From the information gathered,there is no indication that there is a need for significant change to the design of road markings or road signs,since the current standards are reasonably well defined in most jurisdictions.Concerns raised by OEMs predominantly make mention of the degraded state o
83、f road markings and road signs,and not their design.3.3.2 Lane markings for positioning It is possible that as vehicles increasingly use other forms of digital infrastructure and mapping to position and navigate,the issue of road markings may become less critical.However,there is a challenge that si
84、nce many existing advanced driver assistance systems(ADAS)rely on road markings,there may continue to be a reliance on them for some time.An alternative has been trialled on some automated low-speed shuttles,which needed to consider reduced availability of GNSS signals in mountainous areas and tunne
85、ls,and also with reduced functionality of sensors under adverse weather conditions such as fog and snow accumulation etc.Magnetic markers and magnetic-induction lines were installed for vehicles to follow and to identify their location along sections of defined driving routes.Although this approach
86、requires additional investment to provide the dedicated infrastructure,both in the road and vehicles,it enabled vehicles to reliably locate and position themselves along equipped sections.As a result,standards for magnetic markers and magnetic-induction lines are being examined for some automated dr
87、iving technologies to be realized in the early stages in Japan.AUTOMATED VEHICLES 2021R03EN 14 3.4 NEED FOR LANDMARKS Road-side units(RSUs)can also be used to provide landmark information.Static or electronic signs with landmark information,such as QR Codes,WIFI or DSRC,could be provided to assist a
88、utomated vehicles.Several countries such as Germany and China are designing new systems utilising landmarks.Some AV systems,such as low speed AV shuttles,may have limited localization capabilities and may use sensor data in combination with high-definition maps to determine the exact positioning alo
89、ng a route(i.e.proximity to a landmark).Buildings,utility poles,signs or other distinctive landscape features may be used as landmarks or mounting locations for positioning devices so long as they are unlikely to be obstructed by changing environmental conditions 4.3.5 ROAD GEOMETRY,INFRASTRUCTURE D
90、ESIGN 3.5.1 Sign placement An important factor for TSR systems is the location,proximity,and applicability of signs from where they can be viewed the TSR system needs to read only those signs that apply to the vehicle.Some field tests have found that TSR systems can read road signs which are not app
91、licable to the road or lane where the AV may be travelling,such as speed limit signs applying to a separate parallel roadway.Attempts should be made to ensure that placement minimises confusion,both for automated and human driven vehicles.Variation in the position of road signs can lead to them bein
92、g unreadable due to the distance from roadside and angle of the sign face in relation to the traffic direction.Inadequate maintenance that results in signs being dirty,misaligned or having reduced retro-reflectivity can also exacerbate non-ideal positioning.Magnetic markersMagnetic-induction linesse
93、nsorlineMagnetic markerMagnetic MarkerSensorFigure 9:Magnetic markers and magnetic-induction lines(Source:NILIM,MLIT,Japan)Figure 10:Example of automated low-speed shuttle landmark sign Figure 11:Landmark positioning as used in Germany AUTOMATED VEHICLES 2021R03EN 15 3.5.2 Pavement marking and dedic
94、ated lanes During the early stages of automated driving development,nominated routes should attempt to avoiding“dynamic”obstacles such as parked vehicles,pedestrians and/or bicycles.To assist automated driving vehicles in operating smoothly and predictably,measures to suppress on-street parking,pede
95、strian conflicts,and bicycles should be considered.Approaches to clearly define the driving route for automated vehicles(e.g.dedicated pavement marking),and dedicated lanes are under consideration in some jurisdictions.Figure 12:Pavement marking 3.5.3 Universal design needs to be upgraded Universal
96、design requires products to be usable by all people without the need for adaptation or stigmatizing solutions,and this should now also include CV and AV.Universal design principles need to be incorporated not only in vehicle design,but also in the design of related infrastructure,such as electric ve
97、hicle charging stations.One of the more common technology adoptions for CAVs is the provision of necessary information via road-side units.This requires a stable power supply network and a high-capacity communication network along the all interconnected infrastructure in order to provide sufficient
98、and steady services to CAVs.There is a possibility of reducing lane widths in some contexts with the potential for CAV to follow an established trajectory with lower lateral oscillation than human driven vehicles.More lanes could be accommodated within the current infrastructure width,increasing the
99、 capacity of infrastructure at some locations.This reduction of the width could be applied in highly constrained locations such as bridges and tunnels provided that sufficient safety analysis and consideration of all design vehicles(especially for larger vehicles physical widths and swept paths on c
100、urves)has been undertaken.The impacts of some aspects of physical infrastructure design(e.g.road geometry)on vehicle performance could be improved by AV functions.Automation could smoothly recognize horizontal alignments and slopes and adjust performance.Sudden changes in road geometry or width and
101、un-signalized intersections could prove more challenging.AUTOMATED VEHICLES 2021R03EN 16 3.5.4 Crossings At-grade crossings are one of the most crucial physical infrastructure locations that could affect AV and CV design and safe operation.The interaction between pedestrians and AVs will be differen
102、t to interactions with human drivers.One challenge is the determination of responsibility between AVs and pedestrians.Should,for example pedestrians be expected to have an awareness of the presence of AVs,ensuring they dont proceed in front of a CAV at the wrong moment;or should the AV systems be en
103、tirely responsible for detecting pedestrians.Practical limitations of onboard detection systems may result in failure to detect pedestrians under some conditions.Roadside environments can also have objects and furniture that can obscure the view of pedestrians form a vehicle and detection systems ma
104、y not reliably differentiate a human from an object.Uncontrolled or random pedestrian crossing locations:Pedestrians are not a“compliant”or expected part of a system who can be directed and controlled.Pedestrians can be random and chaotic variables for automation to consider which are vulnerable.Ped
105、estrians as potential obstructions on motorways are generally quite rare,although exposure can vary in developing countries,so they still need to be considered for safety.On other road types,the exposure to unexpected crossing behaviour may be greater and the speed environment may be lower but impli
106、cations for automation functions still need to be considered under a wide range of environmental conditions.Zebra crossings:Vehicles are required to stop when pedestrians step onto or are closely approaching a zebra crossing.It may be expected that pedestrians would wait for an approaching vehicle t
107、o slow significantly before crossing,however this may not always be the case.Although the priority at a zebra crossing is better defined,automated functions would still need to consider how to navigate such areas as pedestrians may not perceive the difference between automated and non-automated vehi
108、cles.Signalized crossings:A CAV approaching a signalised crossing needs to be aware of the signal status.It also needs to be recognised that pedestrians may not always wait for the correct signal phase and may cross late or early fail to be clear of the road when the signals change to a vehicle phas
109、e.Signal phase setting such as the length of red times and flashing amber display where used could have effects on CAV operation,and some strategies and design changes may be required to improve clarity.In areas where there is a reasonable expectation of pedestrians crossing the road,whether there i
110、s formalised control or not,road operators need to consider the potential limitations of AV detection systems and also the random behaviours that pedestrians might exhibit.This may involve limiting or clearing some areas of visual obstructions that may obscure the lines of sight near a road where AV
111、s may operate.Consideration may also need to be given to better segregation between pedestrians and vehicles,and channelling pedestrians to controlled or clearer crossing locations to improve mutual visibility and priority for both pedestrians and AVs 3.5.5 Emergency stop areas/hard shoulders The Tr
112、ansport Systems Catapult in the UK(TSC)has assessed the current situation and the change in space of refuge areas and hard shoulders.Traditionally,hard shoulders have been provided along motorways,which provide a continuous strip of hard standing for vehicles to stop in an emergency.AUTOMATED VEHICL
113、ES 2021R03EN 17 However,in more recent years several sections of motorway have been converted to All Lane Running(ALR)to increase their capacity.The ALR system provides safe harbour areas(referred to as emergency refuge areas),which are spaces at intervals of up to every 2.5 km.”Some motorways have
114、been converted to Dynamic Hard Shoulder Running,which involves retaining the solid white line to indicate the presence of a hard shoulder and opening the hard shoulder to general traffic or transit buses during busy times via signs and signals.Figure 13:Example of continuous hard shoulder(left)and E
115、mergency Refuge Area(right)3.5.6 Different factors why the AV should stop in a secure area In full motorway/highway pilot mode,vehicles will be travelling at high speeds with the human driver disengaged from the driving task.It is possible that the driver is not ready to regain control of the vehicl
116、e before it reaches the end of its operational envelope.This could be due to several reasons,such as:Driver falls asleep,suffers some debilitating incident(e.g.heart attack)or becomes otherwise distracted;AV system malfunction or vehicles experience mechanical problems;Deterioration of environmental
117、 conditions;Detection of incidents ahead,such as disabled vehicles in the carriageway,which the AV is unable to negotiate.In this situation the vehicle will need a safe area to stop and wait for the driver to be ready,or for conditions to improve to the extent where the automated control system is a
118、ble to proceed.3.5.7 Safe harbours or hard shoulders?Studies are needed to determine the most appropriate form of safe harbour for AVs,which could also change over time with changes in automation capabilities and the penetration rates of AVs in the vehicle fleet.The advantage of a continuous hard sh
119、oulder is that there is always somewhere to stop at short notice.A disadvantage is that hard shoulders in highspeed environments are not a safe place to stop,especially for extended periods of time.Vehicles travelling in the nearside lane of the motorway can veer into the hard shoulder due to a lack
120、 of concentration by human drivers.A highly AUTOMATED VEHICLES 2021R03EN 18 undesirable scenario would emerge if a driver has fallen asleep and remains asleep for an extended period,placing the vehicle and occupant in a highly exposed and dangerous situation.Safe harbours may need to be appropriatel
121、y designed and contain enough space for an appropriate number of vehicles to stop,and regular enough along a route so that AVs can access them when required.Provisions could also be considered to prevent misuse of these areas recognizing that AVs could considerably increase the use of such areas.The
122、 frequency and spacing of these harbours is an area that road operators may need to consider.The distance between harbours will depend on the traffic demand and conditions,and also the type of AVs that could be present on the segment of road.The ODD will need to consider the availability of shoulder
123、s or safe harbours and the associated risks,which in turn could influence what level of automation might be achievable over various time horizons.3.5.8 How sensors work in bad weather conditions The weather has impacts on road safety which has been clearly demonstrated by historical data.According t
124、o the US Department of Transportation 5,over 5.89m vehicle crashes occur each year on average,out of which around 1.24m can be attributed to adverse weather conditions including snow,rain,fog and severe wind.The most common weather-related accidents are due to wet pavement and rainy conditions which
125、 account for 76%and 46%of the accidents respectively.For the purpose of this report,bad weather is generally defined as wet pavement,rain/hail,snow,icy pavement,snow/slushy pavement and fog.For AVs as well as for the human driver,the implications of low friction and the impact on the effectiveness o
126、f vehicles sensors are important.3.5.9 Low friction In a wet or icy conditions,or on aged pavement,the grip between a vehicle tires and the road can be substantially reduced.Road operators may need to provide pavement friction information and potentially dangerous weather conditions through live dig
127、ital mapping applications and alerts,although this may be data and time intensive to maintain and keep up to date.AVs needs to be able to assess the road condition and determine how best to manoeuvre during turns,how best to proceed from a stopped position and how to best evaluate the required stopp
128、ing distance.Losing traction when accelerating or having insufficient distance to safely come to a stop can result in potentially unsafe driving outcomes.3.5.10 Sensors AVs typically use a mix of different types of on-board sensor technologies including,LIDAR,RADAR and GNSS to continually update the
129、ir digital awareness and position and to navigate safely in their surrounding environment.Adverse weather can affect the accuracy and reliability of these sensor technologies.For example,rain,dirt,leaves or snow can obstruct camera lenses,and the necessary image processing can be impaired or prevent
130、ed without clear imaging.This system is less useful if the lens is not kept clean at all times.Rainfall and snowfall are interpreted by LIDAR sensors as noise that is superimposed over the background physical environment.Figure 14 demonstrates the distortion of the surrounding environment with a sea
131、 of purple objects projected around the vehicle due to snow.AUTOMATED VEHICLES 2021R03EN 19 Figure 14:LIDAR rendering of a snowfall-numerous of purple cloud point(source:Waymo)A common problem for AVs is designing a system that can operate in bad weather conditions.Falling rain and snow tends to cor
132、rupt sensor measurements,particularly for LIDAR sensors.Very little research has been published on methods to de-noise point clouds which are collected by LIDAR in rainy or snowy weather conditions.In a project 7 from the University of Waterloo in Canada,researchers have identified a method for remo
133、ving snow noise by processing point clouds using a 3D outlier detection algorithm.The method is based on the variation in point cloud density with increasing distance from the sensor,with the goal of removing the noise caused by snow while retaining detail in environmental features.Some companies(su
134、ch as Waymo 6)are also using machine learning in their sensor system to filter out the snow and see just whats on the road,even if there are vehicles parked by the curb.It is not clear whether the car can distinguish lanes,but such system capabilities could be sufficient to avoid collisions.Recent t
135、rials indicate that thorough testing in a range of weather conditions including heavy snowfall,rain,sleet,fog,smoke,dust,high humidity,heavy winds,frozen road,ground water accumulation or flow conditions is recommended 4.Road operators need to understand the limitations of sensor systems and filteri
136、ng capabilities as some conditions may remain difficult to resolve for automation and may require additional physical infrastructure/markers,activities to keep roadways clear of certain obstructions/blockages along with reliable digital mapping to assist during adverse weather conditions.3.6 PLATOON
137、ING:IMPACT ON THE INFRASTRUCTURE 3.6.1 Effect of increased vehicle capacity on roads and bridges Some forecasts indicate the automated platoons of vehicles could increase capacity per lane by a factor of up to 40%when conventional vehicles are excluded from the lane.The shorter the platoon gap and t
138、he greater percentage of penetration,the greater the traffic density and road capacity 8.AUTOMATED VEHICLES 2021R03EN 20 There is concern on long span bridges carrying closely spaced platoons where the live-load positive bending moment and shear forces may be substantially greater 9.There may be a n
139、eed to revisit models that assume dilution of heavy vehicles with light vehicles.Aspects of bridge strength that may need to be reviewed include collision on supports,collision on decks,centrifugal forces on curved decks and braking forces increasing the strength of bridges must be done strategicall
140、y(e.g.connect key freight areas suitable for platooning)due to the high cost.Areas that are not suitable may be geofenced accordingly so that platoons dissolve before entering the restricted zone.In some cases,platooning at distances as close as 4m will not result in damage to existing infrastructur
141、e,however it may be necessary to revisit axle weights and spacing between successive vehicles when developing policy around the minimum platooning gaps 8.Pavement fatigue(i.e.rutting)is also a concern,but this may be mitigated by the natural wander of drivers in Level 1 platoons or programmed random
142、 wander at higher levels of automation 10.Each jurisdiction will need to revisit their road design parameters when determining suitable road segments for platooning.3.6.2 Potential changes to infrastructure design to facilitate platooning Dedicated lanes could enhance platooning safety since the beh
143、aviour of other vehicles would be more predictable,and the platoon speed would be more consistent.Dedicated lanes for platooning could also employ more consistent and advanced lane markings which facilitate lateral control and safe operation at higher levels of automation 8.Establishing dedicated tr
144、uck lanes is a potential way to facilitate platooning with minimal impact on existing infrastructure while also enhancing the safety and efficiency of the existing traffic flow.These efficiency benefits are likely to be significantly reduced unless segregated infrastructure is developed 11.Other hig
145、hway design changes that could help facilitate platooning include extended passing lanes,modified ramp acceleration lanes,ramp metering and wider pavement markings 8.However,where road capacity is limited,road authorities must also consider dedicated lanes for public transit needs and public interes
146、t objectives must be balanced.The most efficient use of existing roadways may employ dynamically operated lanes that are customized to meet local area needs;use of HOV and HOT lanes for platooning could also be considered.Consideration may also need to be given to the lengths over which platooning m
147、ay be effective.For example,highway segments with close interchange or intersection spacing may experience disruptive lane changing or high frequency of changes to platoons that could impact the efficiency of platooning.3.7 CITY PLANNING The prospect of highly automated vehicles presents both challe
148、nges and opportunities with respect to city planning and the way communities grow and develop.The spectrum of private versus shared AV models may have diverse impacts on various community types:wilderness/remote,exurban,suburban,inner ring,urban core.3.7.1 Urban Sprawl versus Densification In the ur
149、ban sprawl model,AVs have the potential to make travel more convenient and comfortable allowing commuters to rest,work or undertake recreational activities and therefore develop an increasing tolerance to longer commute times.AUTOMATED VEHICLES 2021R03EN 21 Conversely in the urban densification mode
150、l,AVs also have the potential to reduce parking needs either through shared mobility models or automated valet service allowing cities to reclaim valuable land used for parking lots and garages and repurpose for housing,pedestrian walkways,cycle paths,green spaces etc.It should be noted that the urb
151、an sprawl and densification are not mutually exclusive and may occur in tandem with growing population needs 12.3.7.2 Public Transit Electric low-speed AV shuttles offer the potential to extend public transit and improve mobility to areas that are not well served by existing systems.Automated buses
152、on transit ways could potentially provide cost-effective alternatives to light rail transit.Automated buses operating in a cooperative platoon would offer rail-like service with narrow lane control and optimized acceleration/deceleration with significantly less capital cost 13.It is recommended that
153、 road authorities incorporate CV/AV impact analysis into their planning decisions for new infrastructure and public transit investments.3.8 MAINTENANCE STRATEGIES 3.8.1 Role of Road Maintenance in Enabling AV Safe Deployment The condition and legibility of road signs can degrade due to various envir
154、onmental factors.In addition to traffic sign harmonization,codes and standards development is needed to baseline inspection and maintenance practices as well as establish acceptable thresholds for the extent that pavement markings and traffic signs can crack,fade and deteriorate without compromising
155、 machine vision system recognition.The poor state of lane markings,and signage may hinder CAV deployment in affected areas 14.In rural depopulated areas in Japan,roads are typically very narrow,and there can be vegetation along the roadside.In the example shown in the figure below,bushes were detect
156、ed as obstacles and the automated vehicle stopped even though it was not a dangerous situation.AUTOMATED VEHICLES 2021R03EN 22 Figure 15:Bushes visible on street side(Source:NILIM,MLIT,Japan)The effectiveness and accuracy of TSR systems is affected by the visibility of traffic control devices which
157、can be influenced by rotation,obstruction,deterioration,vandalism,and theft.In regions with heavy snowfalls during winter,the road width becomes narrower due to piled snow.Thus,2-lane 2-way roads sometimes becomes 1-lane 2-way roads.Under such conditions,automated vehicles must change their trajecto
158、ries,or must cope with a“give-way to oncoming car”situation.Figure 16:Example of foliage blocking a low-speed shuttle landmark 4 AUTOMATED VEHICLES 2021R03EN 23 Figure 16:Road width in snowy conditions(Source:NILIM,MLIT,Japan)It is also important that roads are maintained to be kept clear of natural
159、 objects such as overgrown foliage or snowbanks as these may be mistaken for obstacles by machine vision systems and cause AVs to change their trajectories or stop unexpectedly 11.Landmarks relied on by AV systems(e.g.low speed autonomous shuttles)must also be placed and maintained to be clear from
160、natural obstructions 4.AUTOMATED VEHICLES 2021R03EN 24 Figure 17:Examples of pavement marking and road sign maintenance issues 14 3.8.2 AV Use Cases to Improve Road Maintenance There are also opportunities to use automated vehicles for maintenance use cases such as snowplough platooning as well as a
161、sset monitoring and repair.For example,cameras and machine vision systems on specialized vehicles can be used to audit signage on road networks 15.Publicly owned CAVs may also be able to use their sensors and communication functions to report maintenance issues in real time to road authorities 14.AU
162、TOMATED VEHICLES 2021R03EN 25 4 DIGITAL INFRASTRUCTURE-CONNECTIVITY 4.1 NECESSITY OF CONNECTIVITY“Connectivity”means a connection of a vehicle to an ICT terminal via a network,or to another vehicle.Current automated driving technology is mainly based on the sensing systems within the automated vehic
163、le itself,which has many limitations.For example,vehicles alone are:unable to effectively detect non-linear line-of-sight and blind spots vulnerable to rain,snow,fog,haze and other severe weather unable to effectively detect road ice and adjust turning manoeuvres limited to a short perception distan
164、ce unable to achieve all-weather automated driving.The figures below illustrate examples of the ranges within which sensors detect.Sensor recognition systems function within limits-such as range and resolution of detection-connectivity can supplement these limits,enhancing and extending the capabili
165、ties of the vehicle.Highly automated driving requires data fusion that integrates various data provided by multiple on-board sensors and from digital infrastructure.Figure 18:Limitation of range by sensor detection Figure 19:Information on-board sensors are not able to detect Not detectedFallenobjec
166、tDetectedSensor RangeSensor RangeWallDetectedFallenobjectWallNot detectedSensor RangeWallNot detectedDetected AUTOMATED VEHICLES 2021R03EN 26 Data fusion technology,which is defined as the synergistic knowledge from different sources,can assist in creating the overall understanding of a situation an
167、d could be a potential solution to sensor limitations.Figure 20:Data fusion 16 As shown in the figure above,multiple sensors installed in the vehicle enhance detection ability.In addition,the development and deployment of V2X(vehicle to everything connectivity)technologies and infrastructure provide
168、s vehicles with the ability to communicate with each other,addressing to some extent the problem of insufficient information that may be available from in-vehicle systems.Figure 21:Vehicle to Everything(V2X)17 4.2 USE CASES REQUIRING CONNECTIVITY Use cases requiring connectivity from infrastructure(
169、V2I)should be considered separately for high-speed roads(including expressways,motorways,freeways)and ordinary roads due to the differences in travel speed,the presence/absence of pedestrians,bicycles,and parked vehicles on roadways.AUTOMATED VEHICLES 2021R03EN 27 4.2.1 High-Speed Roads Look Ahead I
170、nformation(LAI)provision service The LAI is the information of anticipated events which cannot be detected by on-board sensors.The LAI includes the information about the road ahead including,for example,accidents,disabled cars,other obstacles,events,major incident,wrong-way driving,etc.The National
171、Institute for Land and Infrastructure Management(NILIM),Ministry of Land,Infrastructure,Transport and Tourism(MLIT)of Japan is conducting public-private joint research for next generation C-ITS.Regarding LAI,the following information provision services are identified to realize automated driving ben
172、efits on expressways.Agencies in other countries are also testing connectivity deployments in their jurisdictions which are discussed included the following sections Road obstacle information Since current onboard sensors can only detect obstacles for up to c100m ahead only,it is difficult for autom
173、ated vehicles to avoid obstacles safely by themselves at highway speeds.With information on obstacles ahead,automated vehicles will be able to change their routes,change lanes,or apply the brakes safely.It also enables the prevention of secondary accidents and the early opening of lanes by quickly i
174、dentifying road obstacles.Some car manufacturers operate emergency call service.For example,when the drivers air bag is inflated,emergency information is sent automatically to their emergency call centre.Availability of this information to road operators allows them to respond to an accident promptl
175、y.Figure 22:Road obstacle information provision service A project on the A2/M2 connected vehicle corridor in the UK tested options for a connected Road Works Warning(RWW)system which communicates information about upcoming roadworks such as the location of roadside workers,the configuration of the w
176、orksite,and speed limit requirements for vehicles as they pass the site.It is expected that a system such as this would use a combination of temporary hardware such as worksite beacons and sensors and permanent infrastructure such as Laneregulationinformation(Sections,lanes,terminationschedule,etc.)
177、Road worksRoad administratorLane regulation informationRoad obstacles information(Dissemination of information from damaged vehicles)Vehiclesandautomotivemanagementcentersdeliver information on airbag operation,etc.to theroadadministratorInformation delivered from broken or damagedvehiclesRoadobstac
178、lesinformation(Suspected obstacles early stage,detail ofobstacles,road section after confirmationbythe roadadministrator)*Ensuring the early safety of whistleblowers*Early response to road obstaclesby automobiles*Early response to road obstaclesby automobilesAir bag operationinformation,etc.Road-to-
179、vehicle communicationRoad-to-vehicle communicationTelemarketing CentreRoad administrator AUTOMATED VEHICLES 2021R03EN 28 roadside communication devices to communicate data about the site conditions to connected vehicles 14.The Work Zone Data Exchange(WZDx)Specification enables road operators to make
180、 harmonized work zone data available for third party use.The intent is to make travel on public roads safer and more efficient through ubiquitous access to data on work zone activity.Specifically,the project aims to provide data on work zones into vehicles to help automated driving systems(ADS)and h
181、uman drivers navigate more safely.Figure 23:Work zone USDOT Project phases(WZDx)The U.S Department of Transportation(U.S.DOT)released a$2.4M Notice of Funding Opportunity(NOFO)for WZDx Demonstration,which closed on 3rd August 2020.18 The purpose of this research program is to increase the safety of
182、the traveling public through the production of consistent public work zone data feeds across jurisdictions.This funding provides public roadway operators the opportunity to make unified work zone data feeds available for use by third parties and collaborate on the WZDx Specification development.Cong
183、estion information Congestion information provision service sends information to upstream vehicles about traffic jams(congested section,lane end of congestion,etc.)on interchanges and other exits.This information can be used by AVs to enable them to change lanes in advance,either to join a queue if
184、they intend to exit at the interchange or select a free-flowing lane if intending to continue beyond a congested exit.Figure 24:Congestion information provision service Congestion at IC exits*Enabling drivers to respond quickly to traffic congestionRoad congestion identifiedby the road administrator
185、道路管理者道路管理者Traffic congestioninformation on IC exits(Congestedareas,end of congestion,etc.)V2IcommunicationRoad Administrator AUTOMATED VEHICLES 2021R03EN 29 Tollgate information Current on-board-sensors are not able to recognize with 100%accuracy which tollgates may be open.Tollgate information prov
186、ision service ensures reliable gate choice and provides a safer and smoother passage through a tollgate plaza by providing operational information on each lane.Figure 25:Tollgate information provision service Merging Support Service For merging,automated vehicles need to detect traffic conditions on
187、 the mainline,find space for merging,and coordinate their velocity.Some urban expressways may have relatively short acceleration lanes that may also have constrained sight lines and limit the ability to manoeuvre in the merge area.In order to support automated vehicles to merge smoothly,NILIM is dev
188、eloping a system through public-private joint research which provides information of traffic conditions on the mainline to merging vehicles so that appropriate space to merge smoothly can be identified as automated vehicles approach by adjusting their speed.Figure 26:Merging support service AutopleX
189、 in the UK is a project which enhances an automated vehicles vision and perception,allowing more efficient and effective traffic merging at difficult lane merge situations and roundabout junctions.The project is investigating the fusion of external vehicle sensor information with map Provision of op
190、eration information on each lane*Realization of safe and smooth driving within tollgate terminalsETCGeneralclosureETCGeneralGeneral道路管理者道路管理者Operational information on each laneV2I communicationRoad AdministratorRoad-to-vehicle communicationVehicle detection sensorRoadside processing unitSensing spe
191、eds and lengths other and so on of main-lane vehiclesGenerating information to be provided to merging ADVsProviding merging vehicles with information,such as driving speeds on main-lane vehicles AUTOMATED VEHICLES 2021R03EN 30 aligned infrastructure-based sensing(sensing all road users)which can be
192、transmitted to AVs in real time,along with the applicable road rules context,via standard Internet of Things(IOT)methods 19.Intersection navigation Left Turn Assist(LTA)An application where alerts are given to the driver as they attempt an unprotected(or filtered)left turn across traffic(or right tu
193、rn across traffic),to help them avoid crashes with conflicting traffic from the opposite direction.Figure 27:Left Turn Assist(LTA)Vehicle Turning Right in Front of Bus Warning An application that warns transit bus operators of the presence of vehicles attempting to cross in front of the bus to make
194、a turn as the bus departs from a stop.AUTOMATED VEHICLES 2021R03EN 31 Figure 28:Vehicle Turning Right in Front of Bus Warning Red Light Violation Warning(RLVW)An application that broadcasts signal phase and timing(SPaT)and other data to an in-vehicle device,allowing warnings for impending red light
195、violations.Figure 29:Red Light Violation Warning(RLVW)Spot Weather Impact Warning(SWIW)An application that warns drivers of local hazardous weather conditions by relaying management center and other weather data to roadside equipment,which then re-broadcasts to nearby vehicles.AUTOMATED VEHICLES 202
196、1R03EN 32 Figure 30:Spot Weather Impact Warning(SWIW)Queue Warning(Q-WARN)An application that aims to provide drivers timely warnings of existing and impending queues.Figure 31:Queue Warning(Q-WARN)4.2.2 Ordinary roads Providing information on traffic lights through connectivity has been considered
197、at SIP-adus(the Cross-Ministerial Strategic Innovation Promotion Program:Innovation of Automated Driving for Universal Services)in Japan.Communication of traffic light sates and impending changes is considered AUTOMATED VEHICLES 2021R03EN 33 superior to conventional on-board cameras since detecting
198、signal states at 100%accuracy has been demonstrated to be difficult under various conditions,including backlight.Figure 32:ITS information services at an intersection 20 4.3 REQUIREMENTS FOR CONNECTIVITY It is necessary to determine which communication methods are to be used from the available metho
199、ds as the requirements for connectivity differ for each use case.The following criteria can have different performance requirements for use cases that utilise connectivity:Certainty:Whether information is communicated with full precision Latency:Delay in sending and receiving communication Data rate
200、(MBps)For example,in merging support,an automated vehicle on the mainline may be required to accelerate or decelerate immediately after acquiring information,and thus,a fit-for-purpose communication method with low latency is necessary.In this example the following steps may be involved to illustrat
201、e the steps in the process and the need for high speed communications.(i)Sensors detect traffic conditions on the mainline.(ii)The local server quickly generates merging support information.Information is processed locally as quickly as possible to enable AVs to adjust their speeds.Edge processing,w
202、hich can include the execution of aggregation,data manipulation,bandwidth reduction and other logic directly on an IoT sensor or device,may be used because latency of the system is very important.Edge processing is the on-site processing of data rather than sending it to the cloud or a centralised p
203、rocessing server.Edge processing can reduce the amount time taken to provide a processed“answer”,and also reduce the amount of data transmitted by locally calculating the and providing only the required information.(iii)The roadside antenna provides information about passing time of mainline vehicle
204、s at the merging point to AVs with low latency and reliable communication methods.(iv)On receiving the information,AVs adjust their speeds before entering the mainline to enter the gap in mainline traffic smoothly.AUTOMATED VEHICLES 2021R03EN 34 Figure 33:Concept of merging support service In additi
205、on,road agencies need to take into consideration the costs to deploy connectivity within their jurisdiction including initial investment cost,communication fees and operating cost,and who will ultimately bear the cost.4.4 COMMUNICATION MEASURES The communication methods used for C-ITS are classified
206、 as follows in the TF B.1 report 21.Short Range Communications:5.8 GHz DSRC,5.9 GHz DSRC or ITS G5 and incoming C-V2X PC5.The attributes of this category are the short range(geographical distance)it covers,low latency,capability for two-way communication and small data packet sizes delivered.Long Ra
207、nge Communications:Cellular networks including UMTS(3G),LTE(4G)and incoming 5G.The attributes are the long-range scope,low to medium latency,two-way communications,and larger data packet sizes.Wide Area Broadcast:Digital radio(e.g.DAB+)and analogue radio.The attributes are the long range,medium to h
208、igh latency,limitation to one-way communication and mid-sized data packets.For the merging support system explained above(Figure 33:Concept of merging support service),ETC2.0 is adopted in Japan to develop a prototype.ETC2.0 was adopted as the communication method because the system requires low lat
209、ency and certainty in communication.4.4.1 Future communication method Using 5G technology to achieve collision avoidance and automated driving is similar to using Cellular-V2X,DSRC,and other methods.At this stage it is expected that 5G will have shorter delays and higher reliability.5G will enable h
210、igh-capacity data transmission facilitating detailed 3D maps and also the potential to provide remote control of automatic vehicles where authorised.There are several advanced applications using 5G technology that have been promoted in following areas 22:Automated Driving:the traffic information upd
211、ates could reduce to less than a minute.(iii)Delivering information on such matters as the timing of passing of main-lane vehicles at the merging point(i)Detecting main-lane traffic conditions(iv)ADVs adjusted their speeds before joining the main-lane trafficSensor(ii)Generating information locallyL
212、ocal serverRoadside antenna6 0 k m/h,6 0 k m/h,5 m5 m6 0 k m/h,6 0 k m/h,5 m5 m6 0 k m/h,6 0 k m/h,6 m6 m6 0 k m/h,6 0 k m/h,5 m5 mMerging point AUTOMATED VEHICLES 2021R03EN 35 Figure 34:Automated Driving 22 Platooning:Vehicles operate at very close spacing to save fuel and improve the efficiency of
213、 cargo transportation requiring high speed and reliable data transmissions.Figure 35:Platooning 22 Remote Driving:when the E2E delay is controlled within 10ms,the braking distance generated by the remote emergency braking at a speed of 90 kilometres per hour does not exceed 25 cm.Figure 36:Remote Dr
214、iving 22 5G-V2X realized standardization in 3GPP in 2020,which will be followed by a period of a few years of commercial development.Industry demand plays an important role in 5G technology research and development processes and requirements.Internet of Vehicles has become an important application s
215、cenario for 5G.5G communication technology considers the demand of the automobile and transport industry,and high reliability and low delay has become one of the three major application scenarios defined by ITU and one of the four major application scenarios determined by China.AUTOMATED VEHICLES 20
216、21R03EN 36 5 DIGITAL INFRASTRUCTURE DIGITAL MAPS AND POSITIONING 5.1 INTRODUCTION 5.1.1 What is a digital map?A digital map is a digital representation of a physical environment or asset,combining graphical elements and electronic information to form a virtual representation.They rely on geospatial
217、information information that has a geographic or locational component-to locate attributes and assets.Unlike a physical map that is fixed once created,a digital map can be updated in real time.Digital maps are built up in layers,using Geographical Information Systems(GIS),where each layer is dedicat
218、ed to a type of information,such as roads or contours.Users can choose which layers are of interest and effectively remove unwanted detail or information to enable them to better read the maps.As the maps are virtual and electronic,they can be read at any scale.Digital mapping can be obtained from N
219、ational organisations such as Ordnance Survey(UK),Institut Gographique National(France),National Geographic Information Institute(NGII)Korea.These digital maps can be considered as the“base maps”and are similar to the more traditional paper map,showing a digital representation of the physical enviro
220、nment.These base maps are useful for wayfinding and navigating from A to B.Commercially available navigation systems used in vehicles(e.g.satellite navigation or satnav systems)rely on digital maps to calculate the optimum route to travel;various parameters such as“fastest route”or“avoid tolls”can b
221、e programmed in to these systems.A more sophisticated level of mapping is the high definition(HD)map which includes additional data such as lane markings,signs,speed restrictions and road rules 23.Current generation in-vehicle navigation systems use HD maps and live traffic information,including soc
222、ial media content,to predict the optimum route.Even these systems rely on a human driver to correct for any discrepancies within the data set,road network operators do not accept liability for errors 24.5.1.2 Why digital maps matter to AV The safe and efficient operation of an AV relies on:Navigatio
223、n understanding in real time where the vehicle is in relation to its intended destination Road positioning guiding the vehicle on the road and in relation to lanes Driving in traffic understanding the relative position of other vehicles and road users and driving appropriately to avoid collisions.Wh
224、ereas sensors provide real-time visibility in the vehicles immediate proximity,maps confer a wider view that allows vehicles to anticipate potential critical situations from the surrounding environment.The real challenge for HD map generation is the current level of fragmentation in the automotive i
225、ndustry together with the lack of standards.The concept of a Local Dynamic Map proposed by Shimada et al 25 illustrates the various types and levels of data and information that connected and autonomous vehicles may require for safe operation.AUTOMATED VEHICLES 2021R03EN 37 Figure 37 Level model of
226、a Local Dynamic Map 25 Navigation (Types 1&2)the operation of an AV relies on the ability of the vehicles control systems to safely navigate from origin to destination on a road whilst interacting with road infrastructure,other vehicles and road users.Satnav systems rely on digital maps to plot a ro
227、ute and then to track the location of the vehicle in transit,checking that it remains on the desired route and re-routing when required due to a reported incident for example,or where the vehicle takes a wrong turn.In this navigational process,the vehicle relies on GNSS(Global Navigation Satellite S
228、ystems)to match the vehicles position to the digital map.Many navigation system providers update their maps using data from vehicles using their application via Application Program Interfaces(APIs).One outcome from this approach is that different providers may use different variations of the base ma
229、pping,losing a“single source of truth”.Road positioning (Types 2&3)the safe and efficient operation of an AV requires the vehicle(and its control systems)to understand where it is and to manoeuvre safely using on-board sensors to detect the environment and“see”features such as road marking and signs
230、.An AV uses a variety of systems to understand where it is in relation to the road sensors such as Lidar,digital maps,magnetic guidelines/markers,and Global Navigation Satellite Systems(GNSS)are used in various combinations.Driving in traffic (Types 3&4 data)safely negotiating traffic and other road
231、 users relies on a sophisticated array and combination of on-board sensors to understand a highly dynamic and complex environment rather than a digital map which is relatively static.Some advanced driver assistance systems(ADAS)and Highly Automated Driving systems rely on mapping to augment the info
232、rmation from their sensors 26.It is unlikely that current technology will enable fully automated driving without using data from digital maps to support or supplement on-board sensors and positioning systems this will increase reliance on connectivity to facilitate V2I communications 23.AUTOMATED VE
233、HICLES 2021R03EN 38 5.2 DIGITAL MAP STANDARDS 5.2.1 Base Map The base maps are those provided by the National or State mapping organisations including Ordnance Survey(OS)in the UK,Institut Gographique National(IGN)in France,National Geographic Information Institute(NGII)Korea.Users can subscribe for
234、 automatic updates to the base map.Base maps themselves are not useable by AVs for navigation unlike a human map reader and rely on their conversion to the relevant digital standard that the AV can“read”and use.Base maps are used by commercial organisations(such as navigation and wayfinding provider
235、s)to produce their own mapping systems for use in vehicles.Different standards for digital mapping are used in different countries the European Commissions INSPIRE programme aims to create standardization for mapping 27.5.2.2 Ownership Taking the UK example again,the ownership of the base map belong
236、s to OS.5.2.3 Changes to the base map Permanent changes are provided by the originator of the map,as mentioned above this can be accessed via subscription.Dynamic or temporary changes can have a significant impact on the operation of AVs,such as road works,changes in lane use(e.g.a bus lane)etc.sinc
237、e they are not included in the changes to the base map and require either a Digital Twin or a High Definition map.An important consideration for road operators and AV users/operators is the availability of proposed changes to road layouts and how quickly they can be implemented to the map so that th
238、e AV recognizes the change as soon as it happens 28.Most current navigation systems do not include automatic updates to maps which can be frustrating for human drivers but not generally unsafe,however real time synchronisation between digital and physical information is essential for the safe operat
239、ion of AVs.The European Commission TN-ITS INSPIRE 29 programme aims to provide a platform to enable road network operators to publish changes to their networks so that map makers and service providers can update their maps and make them available to AVs.5.3 DIGITAL TWIN 5.3.1 Definition A digital tw
240、in is“a realistic digital representation of assets,processes or systems in the built or natural environment“30.In the context of a road network,a digital twin is built upon a digital model of the road and all its attributes signs,markings,structures,drainage,fences,etc.Rather than being a static mod
241、el,a digital twin is linked to the physical asset using live data flows from sensors which input current performance data from the physical twin and feedback into the physical twin via real-time control.Therefore,the digital and physical are twinned in real time.Digital twins can exist at any scale,
242、from a street to a city to a national road network.Digital twins are useful for road network operators when designing,constructing,and maintaining their networks,however it is unrealistic to expect digital twins to be created as a retrospective exercise;the cost of digitising existing networks would
243、 be prohibitive.As the use of digital tools becomes commonplace new and upgraded roads will increasingly have a digital twin.AUTOMATED VEHICLES 2021R03EN 39 5.3.2 HD Maps HD(High Definition)maps are like digital twins in that they contain much more information than a standard digital map lanes,signs
244、,road markings,speed limits and other traffic restrictions.HD maps make available layers of information and detail to support and enable AV operations.Commercial organisations have created products to develop and sell HD mapping.The Austroads project team was unable to find commercially available HD
245、 maps of Australia or New Zealand in 2019 23.The researchers concluded that the lack of commercially available maps was likely due to a lack of market demand.The use of HD maps and digital twins is an emerging area and the standards and responsibilities for generation,owning,updating,and publishing
246、are also emerging.5.3.3 Interaction with AV A digital twin will improve the reliability of AV operation by enabling a digital interaction between the information contained within the twin and the AV,rather than relying on sensors to read,process and understand information on physical signs and marki
247、ngs.As the penetration of AVs increases towards 100%,the role of signs and markings for traffic control becomes increasingly redundant.This does not,however,apply to pedestrians,cyclists,scooters,and other modes and would require further investigation into the management and operation of a multi-mod
248、al transport network with varying levels of automation across modes.5.4 POSITIONING 5.4.1 Satellite positioning Global Navigation Satellite Systems(GNSS)are used to determine the location of a vehicle fitted with an appropriate receiver,using signals from satellites.Satellite Navigation has a centra
249、l role in the technology-mix that is currently needed to perform the level of positioning required by automated vehicles;in a recent inquiry made in EU among pilot projects,it resulted in all of them are making use of GNSS as well as fusing GNSS positioning with several other on-board sensors.Since
250、GNSS,as well as other on-board sensors,is affected by external conditions,it is essential that a high quality GNSS is used;this requires optimisation of some key parameters such as availability,integrity,accuracy,sensitivity and robustness to interferences by exploiting all the new GNSS signals/serv
251、ices and recent receivers development(e.g.dual frequency).GNSS is an established technology providing absolute positioning,irrespective of weather conditions.GNSS is independent of any sensors based on perception and may resolve any ambiguities from sensors.It provides accurate timing information,wh
252、ich is needed for sensor fusion,to synchronize the output data of the on-board sensors.Although GNSS could function stand-alone for the less automated functions,there is a continuous influence of environmental conditions and local errors.Therefore,GNSS working in a complementary and interoperable ma
253、nner with other automotive technologies is essential in all the levels of automation defined by the industry.Space Based Augmentation Systems(SBAS)are also used to deliver the positioning requirements for AV 28.AUTOMATED VEHICLES 2021R03EN 40 Galileo In Europe,policy makers understood that Galileo,t
254、he European GNSS,could better support automated vehicles for cooperation between vehicles and for better positioning.The 2016 Declaration of Amsterdam worked to address these barriers and developed a shared European strategy on Connected and Automated Driving.Most recently,the European Parliament ad
255、opted the European Strategy on Cooperative Intelligent Transport Systems(C-ITS)in March 2018 in which Galileo also has a critical role.Australia&New Zealand Free access to an SBAS is not available in Australia nor New Zealand 28 which has the potential to be a barrier to the operation on AVs.The sol
256、ution could be a hybrid system using GNSS and other,ground based,positioning technologies.The result would be a requirement for different hardware on vehicles in these markets which may not be commercially viable.5.4.2 Level of accuracy required The level of accuracy required for automated driving d
257、epends on the level of automation:LEVEL 2-Partial automation:The positioning requirement of basic ADAS functions is at metre-level accuracy(in the range of 2 10m)The GNSS solution meets these requirements complemented by inertial navigation,odometry,and dead reckoning.Type of GNSS solution:normally
258、single frequency receiver(recently multi-GNSS)LEVEL 3-Conditional automation:The positioning requirement of Advanced Cruise Control functions is at half metre-level accuracy(in the range of 40 50cm)The GNSS solution meets these requirements complemented by radar and ultrasound sensors,eHorizon,Compu
259、ter Vision and Simultaneous Localization and Mapping(SLAM)Type of GNSS solution:from single to double frequency,more accurate code measurement augmented by Differential GNSS(DGNSS)and SBAS.LEVEL 4 High automation:The positioning requirement of Autonomous Driving functions is at decimetre-level accur
260、acy(in the range of 25 45cm)The GNSS solution meets these requirements complemented by similar sensors of Level 3 and by 3D Maps,V2V and V2I communication.Type of GNSS solution:from double frequency to triple frequency,high accuracy carrier phase measurements including Real-Time Kinematic(RTK)or Pre
261、cise Point Positioning(PPP),or a combination of both,and authentication encryption LEVEL 5 Full automation:The positioning requirement of fully Driverless functions is at decimetre/cm-level accuracy The GNSS solution meets these requirements complemented by similar sensors of Level 4,and by machine
262、learning and artificial intelligence.AUTOMATED VEHICLES 2021R03EN 41 Type of GNSS solution:as level 4,but guaranteeing more robustness from multiple frequency,enhanced authentication,better availability and very efficient high accuracy/high integrity solution.5.4.3 SF-PPP Single Frequency Precise Po
263、int Positioning(SF-PPP)uses a low-cost receiver with a single frequency,single antenna and single GNSS constellation to provide greater levels of positional accuracy for AV.This has been shown to achieve accuracy of 50cm 31 5.5 SATELLITE BLIND SPOTS GNSS positioning relies on line-of-sight from the
264、vehicle to the satellite(s),therefore when a vehicle is within a tunnel the link to the satellite is lost and the vehicle is no longer able to communicate and establish its position.There are various technical solutions designed to overcome the problems within tunnels to maintain positioning informa
265、tion whilst the vehicle is out of sight of satellites.Thus,GNSS is normally assembled with an Inertial Navigation System(INS),map-matching and dead reckoning techniques.The navigation in tunnels is also supported by the information provided by radar,video sensors or differential corrections from ter
266、restrial stations technologies.The same problem occurs in some city situations where high-rise buildings create“canyons”where satellite signals cannot reach vehicle-based receivers,driving within multi-story car parks or basement car parks presents the same problem.In this case,the use of multifrequ
267、ency GNSS signals can be an appropriate solution to substantially reduce the consequences of reflection in buildings(the so called,“multipath effect”)that distorts the positioning calculation.AUTOMATED VEHICLES 2021R03EN 42 6 DIGITAL INFRASTRUCTURE DATA ISSUES1 COMMON TO CONNECTIVITY AND DIGITAL MAP
268、S Digital Infrastructure is the digital representation of the road environment required by Automated Driving Systems.It includes the connectivity between the vehicle and the road infrastructure,the positioning infrastructure,digital maps and links to advanced Road Management Systems.32 The role of d
269、igital infrastructure is to provide:Support for internal procedures Services to external users,including traveler information services Support to connected and automated vehicles.It is expected that automated vehicles as well as automated functions supporting in-vehicle ADAS will need to be supporte
270、d by additional digital data and services provided by Road Network Operators.In this respect it is important to understand the basic logic of ADAS functionalities.A vehicle uses data from several in-vehicle sensors(including cameras,lidar and radar)and merges them with external data,e.g.map data pro
271、vided by a map-service-provider or data received via C-ITS interfaces.In a so-called“perception phase”all aggregated sensor data are used to build objects resulting in a situational analysis.All data is interpreted to create an“environment-model”which forms a basic requirement for the in-vehicle dec
272、isions to activate or deactivate specific automated functions.If automated functions are activated,the“environmental-model”is aligned with the Operational Design Domain(ODD)where automated functionalities are activated.Figure 38:In-vehicle process to create an environmental model In the context of t
273、he“perception phase”,data from road network operators are highly desired to support automated driving tasks.In addition to the basic data requirements in Europe,a discussion is currently ongoing regarding support for automated in-vehicle functions in case of events,such as poor weather conditions,ro
274、adworks,or other incidents on the road network.The so-called“Infrastructure Support levels for Automated Driving”(ISAD)focus on mixed traffic conditions and define the level of support for ADAS as well as for non-equipped vehicles to enable all vehicles within the road network to be supported with r
275、espective services.These levels can be used to classify the road infrastructure 1 See also PIARC report“Road related data and how to use it”published in December 2020 AUTOMATED VEHICLES 2021R03EN 43 for the supported vehicles,both automated as well as conventional.A“level E road”provides no infrastr
276、ucture support,whilst a“level A road”provides full support for ADAS functions via digital information and by physical infrastructure such as variable message signs(VMS)for conventional vehicles.It is expected that,based on the information received from the road network operator,a vehicle will decide
277、 to activate/deactivate an automated driving function.Figure 39:Levels of the Infrastructure Support for Automated Driving 33 It is evident that not all roads will provide the same range or quality of services to road users,different segments along a road will provide different support levels.Bottle
278、necks and safety critical road stretches and junctions will likely provide greater AV support,while more traditional support will likely be given along rural and remote roads and in peripheral areas.Such scenarios also reflect the current situation,where a better physical infrastructure can be expec
279、ted,for example on motorways than on rural mountainous roads in woodlands.Figure 40 shows,what such an ISAD road classification may look like.The ISAD level information for all road stretches needs to be described to provide the operating environment information to in-vehicle systems and associated
280、service providers.Accordingly,drivers would be informed what services they can expect on what road-links.AUTOMATED VEHICLES 2021R03EN 44 Figure 40:Examples of ISAD levels along the road network 33 The following sections further expand on the detail of the digital elements needed to support ADAS func
281、tions by road network operators,specifically in relation to the processes for data processing and data management.6.1 DATA MANAGEMENT FOR ROAD NETWORK OPERATORS If a road network operator intends to provide digital information into vehicles,proper processes need to be established in addition to the
282、technical systems and equipment required.PIARC Task Force B.1 published a technical report on“Big Data for Road Network Operations”21,describing the framework for big data management and usage based on global best practice.Specifically,big data management deals with capabilities to process big amoun
283、ts of data and to perform proper data analytics using Artificial Intelligence(AI)and machine learning techniques.To facilitate the setup of a digital infrastructure,skills on big data,AI,Internet of Things(IoT)etc.are highly beneficial to road network operators to provide accurate and trusted inform
284、ation into a vehicle.Setting up proper data analytics procedures will result in highly valuable services supporting ADAS functions.The provision of services with high quality timely content,with accurate positioning information(even with lane specific detail)will undoubtably be used in vehicle syste
285、ms.6.1.1 Requirements on data There are mainly two kinds of data to be used by automated vehicles,static and real-time.Static data provides a general description of the road characteristics as well as the legal requirements,(e.g.road rules and general user priorities for making decisions).Real-time
286、data provides accurate information on current road conditions,incidents as well as current active regulations(e.g.variable speed limits or dynamic lane restrictions).Known quality”is a fundamental requirement for both types of data.Smart data is defined as structured accurate data with known benefit
287、s that serves specific use-cases.The ownership and basic resources of data used to support ADAS functions need to be resolved.It is important to know where data comes from and who owns data provided.Only if the resource is known and validated can data be used for digital infrastructure support.AUTOM
288、ATED VEHICLES 2021R03EN 45 For the data itself a structured dataset is of high importance,where raw data are structured in a way that spurious outliers are eliminated.False data needs to be eliminated,and only trustworthy data used in the dataset that forms the basis for RNO services supporting ADAS
289、 functions.The purpose of original data collection also needs to be understood,especially when it is possible that collected data could be reused for other purposes,e.g.to support ADAS functions.Data collected from a road junction(i.e.intersection)for traffic planning issues may have an inherent agg
290、regation level within the collection process which cannot be easily disaggregated.The quality of such data may not suit the needs for other uses that require more disaggregated information(e.g.where lane specific information is needed).As a result,data that has high quality for one purpose might hav
291、e low or even no quality for another use case.For the support of ADAS functions,it is important to be clear on the potential utility of pre-existing data,especially regarding safety,efficiency,environmental guidance,or law enforcement.For the reasons outlined,a corresponding and clear explanation of
292、 datasets is needed,referred to as metadata-information.Metadata ideally describes the content and structure of data as well as the primary purpose for which data has been collected.In this regard the European Commission has published the DCAT-AP(Data Catalogue Vocabulary Application Profile)specifi
293、cation for Metadata 34,which provides guidance to describe datasets in a harmonised structure.That specification is used as basis for data access at European Access Points for transport related data 35.6.1.2 Data processing and data management Data is the basis to enable new services,such as those f
294、or automated vehicle support services.However,raw data has only a marginal benefit or limited useful value by itself.Usually data needs to be processed to create new information or combined with other complementary data to improve existing services.This service generation by a road network operator
295、level is done in integrated,holistic traffic management centres.Therefore,it is highly important to set up internal processes as well as technical expertise to process data in real time to enable highly reliable and quality support to ADAS functions within equipped vehicles.In setting up proper data
296、 processing mechanisms it is important that the resulting services will not only serve automated vehicles,but various interfaces to internal and external stakeholders that need to benefit from improved service quality.These services with improved quality will not only support internal asset manageme
297、nt and planning issues but also assist external stakeholders that can benefit via a broad range of services,as shown in Figure 41.This provision of services via different interfaces needs to be supported by proper data management structures where both internal and external users need to have access
298、to the same data stack through appropriate agreements and controls.Such infrastructure needs to avoid the duplication of data or the provision of similar data with opposing content and with different quality.AUTOMATED VEHICLES 2021R03EN 46 Figure 41:Stakeholders in RNO data exchange 36 Within Figure
299、 41 the focus is on data and service provision from road network authorities towards external stakeholders,including connected and automated vehicles.The traffic management centre(TMC)(or traffic control centre(TCC)as shown in the figure),operated by a road network operator,processes all data and pr
300、ovides access to the data as well as to improved services via different interfaces.All external stakeholders are served by the same,central system.In this example,data can be provided via National Access Points(NAP)(Option 1),which is a European obligation to be set up in EU-countries.ITS Service Pr
301、oviders can access this data via the NAP.,Alternatively,the NAP may only provide a catalogue type service based on metadata which re-routes the request to the road network authority,in which case data access could be given directly by the road network operator(Option 2).An ITS Service Provider could
302、 then use the data received for their own services,e.g.traveller information or navigation services.In addition,the TMC(or TCC)can also exchange data and services with connected vehicles using their own C-ITS infrastructure.C-ITS services can rely on direct short-range communication or via cloud ser
303、vices for long-range communication using cellular networks as an example.It is important to reiterate that whatever interface is used,data exchanged or transmitted always needs to come from a central data management structure set up at the TMC(or TCC)of the road network operator.It is also expected
304、that data collected by vehicles will also be accessible for use by road network operators.Single vehicles could serve as moving sensors improving data and service quality of the road network operator.This data exchange and access to data opens several additional questions regarding data access in ge
305、neral and specifically on data usage,legal aspects,and cost of data.6.2 DATA ACCESS To enable the proper usage of data for traffic management purposes,as well as for the creation of services to support ADAS functions,access to data is crucial.The use of standardised data exchange interfaces,based on
306、 universal standards,is highly recommended.Such standardised interfaces enable easy access and use of data without intermediate complex conversion processes,as during conversion processes data elements and critical knowledge might be lost.Proprietary interfaces are mainly used to strengthen a direct
307、 dependency between a data-provider or component supplier and a data-user.AUTOMATED VEHICLES 2021R03EN 47 However,in a quick moving world the use of standardised interfaces is highly recommended to ensure flexibility in terms of business-relations,even after the installation of components or the est
308、ablishment of cooperation.Additionally,further data providers that contribute to new levels of quality,or the creation of even new services,can easily be added by using a standardised data exchange interface.Alongside the technical aspects of data quality,access and use,several other issues related
309、to business and legal aspects need to be considered,when discussing data access and use.Liability issues,in particular,are developed in Austroads Report(AP-R581-18)Section 2.1 6.2.1 Willingness to give access to data Data has a monetary value.In principle all actors need to be sensible when sharing
310、data and giving access to third parties.Usually,data sharing is based on cooperation models between different actors that clarify the content of data,based on metadata descriptions,and the commercial contract details.Prior to further discussion of cooperation models,data access principles relating t
311、o different actors in the domain of automated driving need to be discussed and agreed.Data owned by road network operators In many cases road network operators are public entities and need to follow national regulations on data access.In many areas,public data are treated as existing for the public
312、good generally resulting in an open data policy.In Europe,for example,several European regulations define the access to data owned by public entities.Sometimes there are only very generic regulations for all kinds of public data,but there are also regulations in the transport sector defining data ca
313、tegories that need to become publicly available.These regulations are aimed at giving access to public data by private companies for the generation of new services.So,usually road network operators are obliged to give access to their data based on a marginal cost principle.This marginal cost princip
314、le can also be used for service provision by the road network operator.Currently most road network operators are not willing to provide any guarantee(thereby limiting their liability)for any supplied data 37.This is especially the case as current legislation for safe road network operations(in most
315、if not all jurisdictions)still considers the physical infrastructure,including signs and road markings,to be the regulatory element.The need for operating a digital infrastructure that provides regulations in a digital machine-readable format,including the digitalisation of the national road traffic
316、 acts,may be required for higher automated in-vehicle functions.In this regard standardisation bodies have already started to work on the management of electronic traffic regulations(METR);liability issues towards road network authorities might arise in the future.Data owned by private actors Privat
317、e actors associated with automated vehicle operations include vehicle manufacturers,telecom operators,digital map providers/operators,and other C-ITS service providers.Usually,private actors are not directly affected by national legislation on data provision.This is due to data and services provided
318、 by private actors being incremental parts of their business development and product offerings.Therefore,private actors are in principle very careful in allowing access to their data.This can prevent competitors from providing the same service quality,allowing a product developer to maintain a legit
319、imate market advantage based on their investment.Few companies are likely to be AUTOMATED VEHICLES 2021R03EN 48 willing to share their data to avoid their competitors gaining insights into their current development status.For automated vehicles,data exchange of specific types between different actor
320、s and vehicles of different brands will be a pre-condition for their successful deployment.Vehicle-to-Vehicle communication,where one vehicle informs the other on possible dangerous situations,is highly relevant and requires a level of collaboration and data exchange.Currently it is not defined what
321、 kind of data will need to be exchanged between different vehicles for a normal driving task,and what data will be generated by in-vehicle systems.For vehicles travelling at high speed in a mixed vehicle environment,vehicle-to-vehicle data exchange will be of high importance to enlarge the operating
322、 horizon of a single vehicle.Current in-vehicle sensors provide an overview of a vehicles surroundings up to approximately 300m ahead.For a vehicle travelling at 120 km/h,it would potentially have a reasonable overview of the upcoming conditions within the next 9 seconds,depending on traffic and env
323、ironmental conditions.The relative shortness of this period might result in sudden uncontrollable movements that could be experienced in non-harmonised traffic flows.Exchanging data between single vehicles would drastically enlarge the potential operating horizon enabling safer vehicle operation and
324、 more efficient traffic flows.Road network operators would benefit from data generated by private actors.If,for example,static geometry data provided by road network operators are not accurate enough,a feedback loop from private actors concerning an alteration to the geometry would help all actors i
325、n the domain.Dynamic data related to efficiency or safety are of high relevance for road network operators.Privately generated data could help operators to improve maintenance processes.Currently data of this kind held by private actors is hardly accessible for reasonable costs,mainly based on busin
326、ess issues,where other private actors might also unfairly benefit from services based on this data.However,where public welfare is concerned,such as the safety of road users,there may be a need for legislation on data access that needs to be followed by private actors.One example is with emergency c
327、alls,where vehicle manufacturers need to provide access to data concerning accidents via telecom operators to emergency services.A similar situation concerning safety is now emerging in connected vehicles.In Europe,legislation exists 38 that identifies categories where private actors need to provide
328、 access to data for free.This includes categories such as the identification of a slippery road surface or the identification of wrong-way drivers.In such cases,private actors are obliged to provide access to their data,if it contributes to the enhancement of road safety.Data owned by individuals In
329、 principle,it is the individual that must decide whether the data generated by their vehicles can be forwarded and used.39 However,usually the individual has only limited influence on their data when signing a contract(e.g.for a vehicle or for a mobile network operator),the rights on data usage is a
330、lso generally handed over to the manufacturer or service provider.Alternatively,when considering connected vehicles where vehicles exchange data with vehicles or infrastructure,it is foreseeable that the vehicle owner may decide whether to provide access to in-vehicle generated data.This principle i
331、s followed in Europe 40,where the owner of the vehicle needs to give their consent to the use of data and to whom it may be given,including the specific purpose AUTOMATED VEHICLES 2021R03EN 49 for the use of data and hence for the identified service.It foreseeable that there will always be an opt-ou
332、t option for end customers and data subjects.Based on this general principle it is important for road network operators to inform travellers about the purpose for which their data will be used.The willingness of travellers to accept a reduced level of privacy will likely increase if a clear reciproc
333、al benefit is perceived.To make the road safer or to improve efficiency might be good argument to convince drivers to provide access to their data.In parallel the supporting services need to be exposed and made understandable for road users.For other service categories(such as law enforcement),national legislation needs to be prepared as most drivers would not willingly volunteer hand data over to