EUCAD 2019: Europe takes the lead
The second European conference on Connected Automated Driving (CAD) “Europe takes the lead” took place in Brussels on 2nd and 3rd April 2019 and was organised by the European Commission with the support of the ARCADE – Aligning Research & Innovation for Connected and Automated Driving in Europe project.
More than 700 international participants – public authorities from the European Commission, Member States, regions and municipalities, industry representative, road and telecom operators, service providers, research organisations universities -gathered in Brussels watched demonstrations, participated in workshops and debated on the future of CAD.
A lively debate on the future of urban mobility brought at the centre of the discussion the crucial role that the automotive suppliers have to deliver: innovative solutions that increase road safety, improvement of traffic flow and reduction of emissions and fuel consumption.
Trust – Safety – Validation – Cooperation: these were the main key topics and priorities emphasised in the opening session in which Jan Carlson, Chairman, President and CEO of Veoneer, Member of CLEPA participated:
“By 2030, the hype currently surrounding autonomous driving has, most probably, materialized. I foresee a gradual acceptance, in the same manner as the PC, the internet and mobile phone hypes. 2030 is only 11 years away, in our industry that means three generations of new car models and a few software updates… Yet I am convinced that even before then, there will be fully automated vehicles in restricted areas”.
A large component of building this trust will come from effective communication between the many moving parts, including vehicle-to-vehicle communication, as well as vehicle-to-infrastructure information sharing. Connectivity – including the roll-out of 5G networks – will be key to unlocking technologies and V2X services, as well as sophisticated planning, roll-out and automation of AI and machine learning at the cloud network edge.
Validation and safety were key discussion topics during the Break out Session on perception systems for cooperative and automotive driving where Alessandro Coda CLEPA CTO contributed as speaker, presenting the automotive supplier prospective.
The session provided an outlook on the current and future challenges for perception systems embedded in and / or communicating with connected and automated vehicles. Fast and reliable perception and Artificial Intelligence are main technical enablers for CAD and great progress has been made in sensing and processing capabilities. However, some situations, like highly dynamic environments in cities or adverse weather and light conditions, are still challenging for perception systems. To address some of these challenges, new systems combinations are emerging: the embedded perception extended with further information provided to the automated vehicles via connectivity services. The session explored as well, how that external information can be used in-vehicle for perception tasks and associated challenges.
Perception systems and AI based systems are by far one of the leading innovation drivers within the automotive industry. This has been proved by the high increase of patents registered during the year. With a total 44% of patent applications registered, the Automotive Industry in Europe is placed at the forefront in leading patenting for those innovations.
Certainly, future technology orientations for CAD perceptions systems will have a strong economic impact on European players.
Because of some situations such as highly dynamic environments in cities or adverse weather and light conditions are still challenging for perception systems thus complex redundant technology is required.
In the coming years, complexity of scene understanding is increasing when we consider urban application of automation. At the same time, requirements to perception systems also increase when it comes to high speed applications. This leads to a huge set of requirements applicable to a high number of perception systems variants needed to cover those applications. Higher robustness, reliability, but also safety computing power and validation are the main impacted requirement areas expected to strongly evolve in the next decade.
Defining the most efficient Sensor sets and Sensing technology mix for automated driving applications is still a major challenge. There is no one standard solution. There will be trade off to be made between the number of sensors we could embedded in the automated vehicle and the realistic computing power that can be embedded in those vehicles for data fusion and scene understanding. This diversity of possible solution also comes with a huge integration challenge, as there is no real standardization for sensors yet established, leading to each sensor describing the vehicle surrounding in its own “dialect”.
AI and Machine Learning are impacting the whole perception chain, from sensors, sensing to scene understanding. We must remain realistic with real capabilities of AI based perception algorithms. Validating that they work with needed accuracy is still an unsolved problem needing a lot of future research and cooperation among research and industry.
Robust and reliable scene understanding under harsh conditions has been improving but still there are major challenges, especially when one tries to extract knowledge about possible intentions of traffic participants to derive from an appropriate driving decision.
Cooperation between research, OEMs, TIERs will be needed to define common and appropriate validation methods, and especially expand the usage of simulation to cover some part of the validation of perception systems.
Perception systems will also need to generate knowledge about what is happening within the vehicle, to generate understanding about the state of the passengers and the engagement level of the driver. This will be needed for all applications where automation is introduced, from passenger cars to buses.
Perception systems need to be combined with highly accurate localization systems and extended by using information provided with low latency and robust precision by connectivity services.
Integration of information from infrastructure or from other vehicle for perception task will be needed to expand perception capabilities of automated vehicles. This comes with a whole new set of requirements towards those external information providers. Concepts to validate and secure the information coming from infrastructure of other vehicles need to be defined to be able to consider connectivity as an additional sensor for perception.
Cybersecurity will become a major challenge also applicable to perception systems. Concepts will be needed to prevent data manipulation leading to taking wrong driving decisions due to corrupted understanding of a traffic scene.
EUCAD2019 exhibition area
CLEPA also supported the EC in building up a dedicated area for EU Research projects on CAD. Examples of latest project results were presented and illustrations made on how the industry will deliver solutions thanks to EU Research funding: Improvement of road infrastructure and network connectivity as well adaptation/ preparation to the future increased level of vehicle automation such as mixed traffic management systems, big data distribution and management, improvement and efficiency on connectivity network, development of human interaction systems with AVs, as well as user acceptance, among other topics. Below, the list of projects involved in the exhibition area during the EUCAD 2019 conference in the Charlemagne building:
- EU 5G Cross Border Corridor
In: CLEPA News, Connectivity & Automation