This multi-stakeholder workshop took place in Torino, Italy on 1-2 October 2019. The workshop put the focus on ODD, costs and benefits of highly automated driving. On Day 1 the workshop discussed the Operational Design Domains, their evolution path and the role they can play in type approval and certification. On Day 2 the workshop examined and discussed the costs and benefits of highly automated driving based upon existing research and projects.The workshop welcomed everyone involved in shaping innovation in the automated driving. Overall, the Workshop attracted in all 37 participants with 15 from industry and the private sector, 17 from public sector and road operators, and 5 representing academia and research.
Download the printable PDF of this page
Maarten Amelink opened the meeting at 13:00 CET.
Introduction to the workshop Angelo Rossini, CEO Aosta Valley Motorway
Workshop welcome was given by Angelo Rossini, CEO of Aosta Valley Motorway, part of the ASTM group, which operates in Northern Italy and Brazil. They have been involved in European projects such as EASYWAY.
Angelo Rossini welcomed the participants and presented a perspective on the importance and priorities for the future of automated vehicles, including user comfort, benefits for the citizens, reduction of accidents, and other social benefits. ODD definitions were also mentioned as critical elements to road operators.
Angelo Rossini is also involved with SATAP autostrada company which operates the A4 motorway Torino – Milano, which is a very busy motorway and now completely renewed.
Torino is the city of Fiat, and has long history in Italy, Lingotto where the original Fiat factory was situated is now completely renewed. Torino wants to lead CAD development in Italy, the important question is the benefits of automation, on accidents and environment, and what Road Operators can do to support CAD.
Introduction of the EU EIP project Roberto Arditi, Coordinator, SINA
Project EU EIP is important which was demonstrated by Mr. Reuters, Deputy Director General of DG MOVE who was present in our Utrecht Workshop. In TEN-T network and ITS networks we need to focus on core networks. We have to contribute to the European policy goals, a problem is that the number of fatalities does not drop any more. We need to build the digitalization layer on top of the physical layer.
Roberto Arditi clarified the importance and role of the European ITS platform.ÂThe European framework helps to join together a harmonized implementation of ITS services.ÂThis collaboration is important to roadway operators.
Goals of the Workshop Maarten Amelink
Maarten gave some practical information and a short overview of the EU EIP Activity 4.2. There is already a draft roadmap which will be updated by the end of 2020. There is a lot of road mapping activity ongoing and cooperation with other activities is very important. The workshop today will contribute to the update of the Road Map.
EC perspectives on ODD Tom Alkim (DG RTD)
Tom Alkim presented on policy objectives from the European Commission, and provided feedback from the recent Connected Automated Driving conference.
The Gartner Hype Cycle demonstrates that more practical characteristics are being addressed now. Tom also highlighted the priorities in the current work cycle and calls in the Horizon 2020 program, plus the STRIA roadmap on Connected Automated Transport.
The second EU CAD Conference took place in April this year in Brussels, the importance of infrastructure was emphasized.gNext week there will be an information day in Brussels where the Commission will explain the latest H2020 calls. The next R&D Framework for 2021 will be called Horizon Europe.
The preparations and stakeholder consultations are already on the way.
The Single Platform for Open Road Testing and Pre-deployment of CAD was established just before the summer. DG MOVE is in charge, with the involvement of DG CONNECT and DG RTD. There is a continuous call for experts, each MS has a seat and can send members .
As regards EU EIP, we said that we need a platform like that in our Athens Workshop. We had good discussions and identified the need to liaise with other stakeholders.
ODD is important: We have to think what we can do to optimise the ODD, to make it continuous, predictable and acceptable.
L3 pilot perspective on ODD Luisa Andreone (CRF)
Luisa Andreone presented highlights on ODD from the OEM perspective. Cooperation and common understanding of priorities and needs are critical to making progress together. iShe presents the L3Pilot view, not of an individual OEM. L3 is testing different automated functions:
The testing will go on for one more year, after that there will be an analysis and impact assessment phase. The ODD discussion started later during the project, it is extremely complex and depends on the use case.
The difference between Automation levels 4 and 5 is fundamental. At level 4 the fall back is human and level 5 it is the vehicle.
The main question is that who is doing what in order to have a continuous and reliable ODD. Cooperation between different stakeholders is clearly needed and hopefully EU EIP can help.
Road operator perspective on ODD Risto Kulmala (Traficon)
Risto Kulmala presented on the road operator perspective of ODD. It included characteristics, costs, risks, and responsibilities of ODD attributes, andÂmessages on deployment and coverage terms.
Risto reported on the work carried out by EU EIP 4.2 and the MANTRA project on ODD. ODD could also be a competitive factor, nobody wants a car which switches continuously between manual and automatic driving modes.
So what is ODD: Static, dynamic, physical, digital. Definition is important as it does not come free. According to the assessments already conducted there are some cost monsters like the safe harbours.
Responsibilities: some are clearly for road operators, some are not so obvious. Some road operators are afraid that automation will increase congestion as the automated vehicles keep longer (and safer) headways.
Transaid perspective Jaap Vreeswijk
Jaap Vreeswijk presented infrastructure assistance in transition areas. How transitions are structured (ToC – Transition of Control), why transition areas are needed, how frequency of transitions are related to level of automation, and areas of recommendation.
The project is dealing with transition areas, some can be very problematic. For instance if a vehicle stops in the roadway, which is a problem especially in high traffic conditions, so safe stopping is required.
The elements analysed by Transaid are:
• System operational performance
• SAE levels
• Infrastructure support level
• Events and other road users
Transaid has also looked into the disengagement reports from US in order to learn about the performance of automated vehicles. There are a lot of open questions. There will be a workshop during ITS world Singapore in October on this topic.
Moderator Maarten Amelink
the panelists: Tom Alkim, Luisa Andreone, Risto Kulmala, Jaap Vreeswijk
Maarten kicked off the panel discussion by asking some questions to the panellist:
Q1: Quite some people and projects are working on ODD – how far are we in reaching a common framework and how important is it to have a common view?
Risto: it is useful to have a common view, road operators have to understand OEMs. OEMs hopefully have also this view, it is good to have these workshops. The workshop organised by Jaap is good opportunity to discuss ODD on global level.
Luisa: The reality is that we are addressing the same issue and use the same language and the word, at least it is a start.
Q2: There is on the one had the general framework for ODD, the way to define it. This can be used to actually define the ODD for a certain service or use case. The capabilities of the systems define the ODD. In what sense is this a static definition or does it develop with the evolution of sensors? … What does this mean for other stakeholders like road operators?
Tom: What we do at the moment is not perfect. We have to go beyond the theoretical framework. ODD is not static, we could visualize a moving car where the shape and size of ODD is changing constantly, contracting and expanding. This dynamic range is where the vehicle is OK and inside its ODD.
Maarten: ODD has dynamic and static part. Risto: in two years a car may not need lane markings. Tom: This will depend on the capability of the vehicle sensors.
Q3: How ODD should be considered in the type approval and certification of highly automated vehicles?
Luisa: We are not there yet, first we need a permission to conduct testing first. But the OEMs want to sell cars, and for type approval we are behind schedule here (we as a community, not one research project like L3Pilot).
Tom: You cannot type approve automated vehicles, we do not know yet how to do it. First we have to develop methodology for the automated capabilities, which could be restricted to certain roads (the case of ODD).
Jaap: General Motors is using ADAS geo-fencing which is kind of ODD. The vehicle is guaranteed to be safe only inside the specific area.
Luisa: We have to work together especially on the dynamic features as much of that is information that can be provided by the Road Operators (e.g congestion). Jaap: We have to look into metrics/units and the thresholds for each parameter.
Risto: Vehicle type approval is problematic as the functionality of vehicles changes all the time because of the (wireless) updates.
Questions from the audience:
Merja Penttinen, VTT Finland: What about the safety? The most risky parts of the network, interurban and rural roads, are not dealt with now
Risto: We start with use cases to be deployed first, and those are deployd either on motorways or on urban streets.
Tom: This is a paradox, there is a trade off between complexity and Velocity. They are different in every environment – urban areas are very complex but velocities are low. Motorways are less complex but velocities are higher. Maybe we need to take a step back, look into the problem first – CAD is not always a solution. Maybe in urban all trips do not have to be taken by a car. Drive less – not driverless!
Johnny Svedlund, Trafikverket Sweden: We have already have ODD for different types of vehicles, i.e. you cannot drive a scooter on the motorway. This is not new for road operators.
Tom: Should we type approve human behaviour?
Walter: There is a lot of work done but still no answers, some stakeholders point out what others should do which is a common problem.
Maarten: Conclusion of this is that each stakeholder should think what he can do for ODD, not just what others can do.
There were three parallel sessions for three Use Case groups based on the Use Cases of the ERTRAC Road Map.
Each moderator presented a short summary reporting on the discussions on the ODD attributes for each Use Case (see the resulting Excel tables):
Download day 1 parallel session proceedings
Opening of the second day by Johnny Svedlund
Roberto Arditi SINA: very happy with the quality as evidenced yesterday, DG MOVE asks to keep the Road Operators involved in CAD
Introduction to impacts, cost and benefits of highly automated vehicles: Magnus Hjälmdahl
Aim of the second day: Examine and discuss the costs and benefits of highly automated driving based upon existing research and projects.
Same groups as yesterday, building on the results. We start with introduction presentations. Plenty of time for the workshop and discussions.
Introduction of the Task two of EU EIP 4.2. Direct and indirect costs, road operator perspective but not only. Final Report due by the end of the year.
Presentation 1: Hendrik Weber: Impact Assessment in AdaptIVe and peek into L3Pilot
Hendrik Weber presented his research on AdaptiVe and in connection with L3Pilot. A safety impact assessment was conducted considering manoeuvre scenarios in which the AVs must perform.
The findings demonstrate how AVs in the given manoeuvre scenarios compare in terms of collision frequency relative to general collision data in Germany.
EuroFOT was a precursor and also did an impact assessment, next after that was Adaptive which ended two years ago. AdaptIVe did also an impact evaluation, and it will be done in the L3Pilot after the pilots.
AdaptIVe focused on Safety impact assessment (one use case) and Environmental impact assessment (all use cases)
Used accident data e.g. GIDAS database and building driving scenarios for the analysis. Using simulation which of course has some limitations. Environmental impact was assessed with the same methodology.
Penetration rate is important but also usage data (if the functionality is actually used).
Conclusion: Automation can have positive effect but that depends on penetration rate/usage.
Next is the L3Pilot where a lot of data will be collected and analysed – again safety and environment, followed by cost-benefit analysis.
Presentation 2: Pirkko Rämä (VTT Principal Scientist): CARTRE, Scenarios and their benefits
(Results of the CARTRE benefit evaluation based on the four future deployment scenarios)
Pirkko Rämä presented the Impacts of automated driving CARTRE: Expert estimate in future deployment scenarios.
A literature-based set of scenarios were applied to a small expert working group to evaluate the impacts to the success of shared mobility relative to the stress level on the role of public authorities.
The findings demonstrate how the KPIs relate to targeted directions in categories (Use and Acceptance, Mobility and Travel Behaviour, Public Health and Safety, Land Use, Economic Benefits for the Society, Costs and Investments for Society). Relationship to ODD was also estimated.
CARTRE – Pirkko was leading the impact assessment part. Limited resources.
Ex-Ante: Long term effects by the decisions done now, socio-economic impacts cannot be estimated.
In CARTRE there were 17 experts. Eight impact areas were analysed. Initial KPI’s coming from the tri-lateral EU-US-Japan cooperation in CAD. Complicated, so scenario-based approach using earlier work was adopted. The four scenarios were
Results are opinions of the experts, not actual measured data: CARTRE also estimated behavioural impact of highway autopilot and urban autopilot.
Also, some preliminary estimates on network effects.
Explanation of the Excel tool by Vish, impacts value – 10 to +10 and weight from 0 to 100, for road operators and others – for different scenarios. Same for the costs, taken from previous work.
Parallel sessions per use case group, same as Day 1. The difference being in these sessions the participants worked with Impacts (Score and Weights for impacts), and costs (Score of costs for Road operators, Weight of costs for Road operators and Score for Others)
Download day 2 parallel session proceedings
4. Feedback and discussion
Short presentations of the groups
5. Conclusions & wrap up day 2, closing of workshop
In his summary Magnus said that we have now a lot of material, will now prepare the final report.
Meeting ends at 14.45